Small update, for anyone who may be curious. I just purchased Mylar in 1 mil and 1/2 mil thicknesses from a
model airplane shop. The 3D printed molds for the diaphragm are still in production, and I'm very curious to see how such a thin material holds up to vacuum forming. I have a feeling I may need to create a positive of the mold based on the 3D printed negative and press the heated mylar between the negative and positive molds, instead of vacuum forming, but we'll see.
For vacuum forming there's an intermediate thing between one-sided and two-sided molding, where you use a one-sided mold and push the plastic most of the way toward it, into the right general shape, and let vacuum take it the rest of the way.
So if you're using a negative mold, you'll have something roughly the positive shape, but smaller, to push the plastic into the mold.
For one-offs or a-few-offs of reasonably large objects, people often do that with their hands in work gloves, gently stretching the plastic toward the parts of the mold that it's hardest for the vacuum to stretch it into. Once the plastic has been stretched some, it's thinner and easier for the vacuum to suck it the rest of the way.
For larger volume things, people will often make a felt-covered thing to push the plastic into the mold in the right places. (Felt covered so that plastic can slide across it and not be cooled much by contact.
In general you do NOT want your molds to have a polished surface, because the plastic may stick, and it's harder to suck the air out everywhere. You want the mold to be rough at a scale significantly smaller than the thickness of the plastic---small enough that the plastic will not mold to it and become rough at that scale, but still large enough that air can work its way out through the tiny gaps the roughness creates between the plastic and the mold.
(Worse case, the plastic may stick to the mold. Normally you heat the plastic enough to be stretchy but not enough to be sticky, but if you misjudge, and have a smooth mold, that can be bad.)
For very very low volume stuff, like one to ten things, people often make molds of some kind of plaster. (Plaster of Paris or something denser and stronger like Hydrocal, or tougher like Water Putty.) That naturally gives a microscopically rough surface and considerable porosity---you can suck the last bits of air right through plaster.) Plaster molds tend to wear out quickly, so one option to make a silicone mold to cast several plaster molds from, if you're making a moderate number of things.
Molding very thin plastic is tricky, because it tends to cool quickly and give you less working time and other forms of slack. For example, if you have a felt-covered thing to press the plastic toward the mold, just contact with the felt may cool it too much. You may need your mold and that other thing (if you use one) to be kept warm so that they don't cool the plastic too much before it conforms to the mold.
(Vacuum forming is easiest with plastics between about one millimeter and and four millimeters thick, where the heated plastic is thick enough to retain most of its heat for a few seconds, but not so thick it's hard for vacuum to stretch it to fit the mold.)
BTW here's a video of me forming plastic over a plaster mold of my face (keep the volume low... there's no useful audio content and you'll mainly hear a vacuum cleaner):
Here's my Instructable on how to make that vacuum former:
https://www.instructables.com/Make-a-good,-cheap,-upgradeable-sheet-plastic-vacu/
Here's my Instructable on how to turn a cheap bike tire pump into a vacuum pump that can pull several times as hard as a vacuum cleaner:
https://www.instructables.com/make-a-manual-vacuum-pump-for-under-$20-by-convert/ And another using a cheap electric car tire inflator:
https://www.instructables.com/convert-a-tire-inflator-type-air-compressor-into-a/
So I press for the source you said you found.
