Simplify VFS initialization

[sosthene-nitrokey]

Can be improved further once trussed-dev/littlefs2#87 is merged.

[robin-nitrokey]

probably debugging artifact?

[robin-nitrokey]

Why not VolatileStorage::new()?

[nitrokey-ci]

No significant changes.

Insignifcant changes

metric	value		change
binary-size-nk3am	1,502,577	✅	-217 (-0.01%)
binary-size-nk3am-test	1,880,394	✅	-119 (-0.01%)
binary-size-nk3xn	539,984	✅	-316 (-0.06%)
binary-size-nk3xn-test	540,080	✅	-316 (-0.06%)
binary-size-nkpk	712,451	✅	-590 (-0.08%)

[robin-nitrokey]

Note sure what causes the binary size change. I cannot reproduce it locally.

[sosthene-nitrokey]

I do not understand the size increase. On the opposite I would have expected a size decrease.

[robin-nitrokey]

My first guess was that maybe the build environment changed and caused the size change, but running the CI for a no-op PR like #240 does not cause a size difference.

[sosthene-nitrokey]

I can reproduce the size change locally.  My first thought was a missing rebase but it's not that.

[robin-nitrokey]

I only checked the output of make bloat locally, not of the binary size. It is quite the coincidence that the volatile storage size is 8192 and the size change is e. g. 8176. It looks like this is just the size of the initializer? Using a MaybeUninit to defer initialization to runtime could help.

[sosthene-nitrokey]

Yes it looks like it was that. It's including the full zeros of the static? That's not good. As a result using MaybeUninit improves the binary size by the couple of relevant bytes.

[daringer]

mmmh, falsely approved although the build fails - can't check it from here