Add MSVC coverage for remaining suspicious asserts

[chiphogg]

I went through every static_assert in our C++ files, and added MSVC
test coverage for any that looked like they might be vulnerable to the
same error as in #363.

I added these test cases one at a time, so I could tell which ones
failed.

• The is_known_to_be_less_than_one "bug" never actually failed,
  whether I called it directly or indirectly. I'm leaving the tests in
  place, and I "fixed" it anyway because it made the function body more
  stylistically consistent.

• The inverse_as/inverse_in bug was indeed another instance of this
  problem. Now we have coverage and a fix.

Fixes #364.

[geoffviola]

The change seems fine. I'm not sure I understand the original issue. The error that I see is

failure was caused by a read of a variable outside its lifetime

But I don't see how that is possible 😕

With the changes to single-file-test.cc, I see the issue in godbolt. So the coverage seems better.

As an aside, can we get constants in the single file deployment? 😄 It doesn't seem to have SPEED_OF_LIGHT.

[chiphogg]

The change seems fine. I'm not sure I understand the original issue. The error that I see is

failure was caused by a read of a variable outside its lifetime

But I don't see how that is possible 😕

I found this super confusing as well. The key is that "outside" in this case probably means "before". It's running "at compile time", but we're using the value of a variable that won't exist until runtime. At least, that's the nearest I can figure.

As an aside, can we get constants in the single file deployment? 😄 It doesn't seem to have SPEED_OF_LIGHT.

Those single-file releases are barebones. Check out the "Pre-built files with all units" box from the single file section of the install instructions. They should all be there! And, they're also in our canonical godbolt link from the README.md.

[hoffbrinkle]

failure was caused by a read of a variable outside its lifetime

But I don't see how that is possible 😕

I found this super confusing as well. The key is that "outside" in this case probably means "before". It's running "at compile time", but we're using the value of a variable that won't exist until runtime. At least, that's the nearest I can figure.

Interesting. Is this an "issue" with MSVC? We're using it at compile time, but are we using any information that isn't available at compile time?

[chiphogg]

failure was caused by a read of a variable outside its lifetime

But I don't see how that is possible 😕

I found this super confusing as well. The key is that "outside" in this case probably means "before". It's running "at compile time", but we're using the value of a variable that won't exist until runtime. At least, that's the nearest I can figure.

Interesting. Is this an "issue" with MSVC? We're using it at compile time, but are we using any information that isn't available at compile time?

Possibly! But I think we'd need to go deep into "language lawyer" territory to definitively say who is right. 😅