[roll] Roll fuchsia [superproject] Roll llvm-project/libc [libc][math] Implement double precision expm1 function correctly rounded for all  rounding modes. (#67048)

Implementing expm1 function for double precision based on exp function
algorithm:

- Reduced x = log2(e) * (hi + mid1 + mid2) + lo, where:
  * hi is an integer
  * mid1 * 2^-6 is an integer
  * mid2 * 2^-12 is an integer
  * |lo| < 2^-13 + 2^-30
- Then exp(x) - 1 = 2^hi * 2^mid1 * 2^mid2 * exp(lo) - 1 ~ 2^hi *
(2^mid1 * 2^mid2 * (1 + lo * P(lo)) - 2^(-hi) )
- We evaluate fast pass with P(lo) is a degree-3 Taylor polynomial of
(e^lo - 1) / lo in double precision
- If the Ziv accuracy test fails, we use degree-6 Taylor polynomial of
(e^lo - 1) / lo in double double precision
- If the Ziv accuracy test still fails, we re-evaluate everything in
128-bit precision.

GitOrigin-RevId: 3e7d99aa57be39bffbc659fcf31e4a0048ecf406
Original-Revision: e04ae1965354d0573affc03b925218916afe99e4
Roller-URL: https://ci.chromium.org/b/8768662457257879121
CQ-Do-Not-Cancel-Tryjobs: true
Original-Reviewed-on: https://fuchsia-review.googlesource.com/c/fuchsia/+/924147
Original-Revision: 2fb4282a5391a88c6cb7b5ab507cd601159e91e6
Change-Id: Icaa4f5058b2b909de41d06ae4c5948fefb3b7b2b