<thread>: Avoid overflow in _To_absolute_time()

[StephanTLavavej]

Fixes #5234.

We were carefully forming decltype(_Now + _Rel_time) _Abs_time, the common type of the given _Rel_time (which could be coarse or fine) and chrono::steady_clock::now() (which is nanoseconds).

In this bug scenario, _Rel_time and therefore the common type are picoseconds.

The problem was that we weren't careful about our _Forever constant. We were using (chrono::steady_clock::time_point::max)(), but then converting it to the common type (implicitly in the subtraction _Forever - _Rel_time, and in the assignment _Abs_time = _Forever). When the common type is finer-grained, this attempts to multiply the stored value, but it's already the maximum, so we overflow.

The fix is to use the common type for _Forever. This stores the maximum rep with the proper period of the common type, so we won't attempt to multiply it further. (When _Rel_time is coarse, the common type is nanoseconds, so there's no change. When _Rel_time is fine, this means that _Forever is effectively "earlier", because INT64_MAX picoseconds from the epoch happens earlier than INT64_MAX nanoseconds. This is exactly what we want, since we're performing our overflow check and clamping with the common type that we're about to return.)

[StephanTLavavej]

I'm mirroring this to the MSVC-internal repo - please notify me if any further changes are pushed.