Windows: Make builds more reproducible

[spycrab]

Hopefully helps us with whitelisting our updater...

[BhaaLseN]

Somewhat sad that this has to be an undocumented flag. At least in Roslyn, theres the documented /deterministic switch and the project option <Deterministic>true</Deterministic>.
But it could be worth a try. +1 from my side unless someone specifically disagrees (or has a good reason to not do this).

[spycrab]

I'm not quite sure yet if this actually helped any. It looks to me as if at least one timestamp remains. Has to be looked into further

[BhaaLseN]

I did a few builds/rebuilds and came out with the same md5sum; even though switched commits between them. So I guess...it kinda works?

[spycrab]

Well it might be at least better than nothing.

[mbc07]

Doesn't seem to work, updater binary still get different hashes after this PR. At least I successfully downloaded 5.0-11044, 5.0-11046 and 5.0-11048 without triggering false positive alerts on Windows Defender, but unsure if the problem was really fixed or if it's just that somebody else already submitted those binaries for analysis...

[BhaaLseN]

We don't bake the sha1 in like we do with the main executable, do we? I'm not quite sure why this happens tho...