I've been doing a bit of research on this topic with a couple of buddies since this news broke, and using my amazing powers of deduction, I've come to the conclusion that with some confidence you can say that the Missingno slots with cries were once indeed filled, though to an unknown degree.
Given the confirmation of Missingno being reserve slots for future Pokemon, we can now say reasonably what the process was for creating a new Pokemon: First, an Index number was defined, along with the most basic bits of their data. Then, Moveset/learnset/stats/etc. are defined and stored with the Pokedex number- these two in combination making a complete Pokemon- and finally of course distributing it through the world in encounters and what have you.
Now, this is still a theory because, well, I didn't develop the game so I couldn't say how this went, and you have to assume what I just said up there is true for the next bit to make sense. Still, with the layout in the Index known to be (mostly) older than the Pokedex layout, as well as the confirmation everyone's buzzing about, I think that is a reasonable enough assumption to make.
Now, this gets interesting for 2 reasons. The first, I'll admit, is stretching a bit, but I think the second is quite hard to argue with in these conditions:
1. There's a distinct possibility "Cactus" made it into the game at some point- Not only does its index slot have a unique cry, on the "Popularity Vote" sheet, it actually trumped a few Pokemon that made it into the final game. Even if "Sleeper" is not Hypno, as is speculated, it still beat Nidoking, Scyther, and Diglett.
2. I will admit I'm not the first person to notice this, as user "Koolboyman" noted this in this very thread 8 years ago, but 5E and 5F fall within a sea of two-stage evolution Pokemon (ranging from Vulpix and Ninetails at 52 and 53 to Jigglypuff and Wigglytuff at 64 and 65, only interrupted by 2 more Missingno that are also likely to be a pair and Dratini/Dragonair who themselves were likely once disconnected from Dragonite) and have defined and similar cries, as common with other Evolutions- 5F's being lower pitched, as is often the case with a larger Pokemon in the second stage.
The conclusion would be that either these Pokemon had data assigned to their Index value but nothing further than that, or that their Pokedex-related info was blanked and recycled for use for a Pokemon later down the line- which would explain why some Pokemon near the bottom of the Index like Oddish and its line ended up so early in the Pokedex. No real way to tell how far it got, and unless someone gets their hands on a Prototype or something, no real way to tell, hence "an unknown degree". All I can say is- there was likely something there at some point. As for slots with no defined cry- far less likely to have ever been anything more than placeholder positions.
As for "why they would just leave cries", I think the best answer is simple: Pokedex data carries a big footprint. It takes up more file size, and in case of some sort of an error, leads to a scenario where a casual playtester or player would likely not notice something is wrong with their game as a Pokemon with a valid sprite, moveset, learnset, etc. appears when it's not supposed to.
On the other hand, an Index Number barely has anything left, with a defined cry being the only tell something was there. A scant presence like that would probably not even be noticed- especially by the same team who failed to notice bugs in the final product as obvious as "Focus Energy does the opposite of what it's supposed to". Even if they did notice that the indices still pointed to some valid data, it probably wouldn't be worth the effort deleting it unless they really wanted the index for something else- after all, an unused town and unused song point to cleanup in this game not being all that thorough anyway.
As far as an idea proposed by (I believe) Crystal_ that the slots used to test various cries- I feel like 5E and 5F's correlation work against that theory, and there would be several easier ways to test them (replace the cries of Pokemon you encounter at the start of the game, or test them with text box events via a debug NPC ala Mr. Chrono in G/S).
Also, the slots with a cry predate Muk- meaning likely they predate Biker teams that give Missingno its form- and I haven't tested it myself, but smart money says trying to load a Pokemon without a defined sprite or stats is probably bad juju.
Given the confirmation of Missingno being reserve slots for future Pokemon, we can now say reasonably what the process was for creating a new Pokemon: First, an Index number was defined, along with the most basic bits of their data. Then, Moveset/learnset/stats/etc. are defined and stored with the Pokedex number- these two in combination making a complete Pokemon- and finally of course distributing it through the world in encounters and what have you.
Now, this is still a theory because, well, I didn't develop the game so I couldn't say how this went, and you have to assume what I just said up there is true for the next bit to make sense. Still, with the layout in the Index known to be (mostly) older than the Pokedex layout, as well as the confirmation everyone's buzzing about, I think that is a reasonable enough assumption to make.
Now, this gets interesting for 2 reasons. The first, I'll admit, is stretching a bit, but I think the second is quite hard to argue with in these conditions:
1. There's a distinct possibility "Cactus" made it into the game at some point- Not only does its index slot have a unique cry, on the "Popularity Vote" sheet, it actually trumped a few Pokemon that made it into the final game. Even if "Sleeper" is not Hypno, as is speculated, it still beat Nidoking, Scyther, and Diglett.
2. I will admit I'm not the first person to notice this, as user "Koolboyman" noted this in this very thread 8 years ago, but 5E and 5F fall within a sea of two-stage evolution Pokemon (ranging from Vulpix and Ninetails at 52 and 53 to Jigglypuff and Wigglytuff at 64 and 65, only interrupted by 2 more Missingno that are also likely to be a pair and Dratini/Dragonair who themselves were likely once disconnected from Dragonite) and have defined and similar cries, as common with other Evolutions- 5F's being lower pitched, as is often the case with a larger Pokemon in the second stage.
The conclusion would be that either these Pokemon had data assigned to their Index value but nothing further than that, or that their Pokedex-related info was blanked and recycled for use for a Pokemon later down the line- which would explain why some Pokemon near the bottom of the Index like Oddish and its line ended up so early in the Pokedex. No real way to tell how far it got, and unless someone gets their hands on a Prototype or something, no real way to tell, hence "an unknown degree". All I can say is- there was likely something there at some point. As for slots with no defined cry- far less likely to have ever been anything more than placeholder positions.
As for "why they would just leave cries", I think the best answer is simple: Pokedex data carries a big footprint. It takes up more file size, and in case of some sort of an error, leads to a scenario where a casual playtester or player would likely not notice something is wrong with their game as a Pokemon with a valid sprite, moveset, learnset, etc. appears when it's not supposed to.
On the other hand, an Index Number barely has anything left, with a defined cry being the only tell something was there. A scant presence like that would probably not even be noticed- especially by the same team who failed to notice bugs in the final product as obvious as "Focus Energy does the opposite of what it's supposed to". Even if they did notice that the indices still pointed to some valid data, it probably wouldn't be worth the effort deleting it unless they really wanted the index for something else- after all, an unused town and unused song point to cleanup in this game not being all that thorough anyway.
As far as an idea proposed by (I believe) Crystal_ that the slots used to test various cries- I feel like 5E and 5F's correlation work against that theory, and there would be several easier ways to test them (replace the cries of Pokemon you encounter at the start of the game, or test them with text box events via a debug NPC ala Mr. Chrono in G/S).
Also, the slots with a cry predate Muk- meaning likely they predate Biker teams that give Missingno its form- and I haven't tested it myself, but smart money says trying to load a Pokemon without a defined sprite or stats is probably bad juju.