[roll] Roll fuchsia [tracefs] Lockless ring buffer

Implements the tracefs ring buffer with a lockless design.

This is based on the design in https://docs.kernel.org/trace/ring-buffer-design.html

The implementation is using Rust, so there are some differences when
dealing with pointers and some other data structure differences.

This implementation takes advantage of only reading full pages of events
when the page is completely committed. By doing this we can handle
writes to the same page being reserved in order, but committed out of
order and only reading the page when all writes are committed.

The performance improvement is

Reading: P99 improved 46.5% from 73.59 us to 39.30 us.
Writing: P99 improved 33.4% from 53.59 us to 35.68 us.

Reading: average improved 75.6% from 32.77 us to 7.98 us.
Writing: average improved 42.7% from 16.92 us to 9.68 us.

Multiply: starnix_core_tests:perf::
Original-Original-Bug: 414642814
Original-Original-Reviewed-on: https://fuchsia-review.googlesource.com/c/fuchsia/+/1585075
Original-Original-Revision: 66694476e90ab059fca4bc6b1196721cdae7d0bc
GitOrigin-RevId: 6e8a5e8891dd4483d96ef22f97a7e3e6a67f88c6
Change-Id: Icdc9c84545d9ab34b2a9d24b2d6f94c98d94bc1a
Reviewed-on: https://fuchsia-review.googlesource.com/c/integration/+/1622161
Cr-Commit-Position: refs/heads/main@{#193247}