timestamp modification system call support

[libumem]

Checklist

Which kind of PR do you create?

• This PR only contains minor fixes.
• This PR contains major feature update.
• This PR introduces a new function/api for Qiling Framework.

Coding convention?

• The new code conforms to Qiling Framework naming convention.
• The imports are arranged properly.
• Essential comments are added.
• The reference of the new code is pointed out.

Extra tests?

• No extra tests are needed for this PR.
• I have added enough tests for this PR.
• Tests will be added after some discussion and review.

Changelog?

• This PR doesn't need to update Changelog.
• Changelog will be updated after some proper review.
• Changelog has been updated in my PR.

Target branch?

• The target branch is dev branch.

One last thing

• I have read the contribution guide

---

This PR adds support for the syscalls:

• utime
• futimesat
• utimes
• utimensat

and adds source code for an example to be merged in the rootfs repo.

[libumem]

Converting to draft until qilingframework/rootfs#37 is handled

[xwings]

Two issues need fixes before merge:

1. times is nullable for utime(2), utimes(2), and utimensat(2). Linux treats a NULL pointer as "set atime/mtime to now" subject to the normal permission rules. This implementation currently either dereferences 0 in do_utime() or returns positive EACCES from ql_syscall_utimensat(), which is not a syscall error under Qiling return conventions. Please handle times == 0 by setting both timestamps to now and return negative errno values on failure.

2. futimesat(2) takes struct timeval times[2], while utimensat(2) takes struct timespec times[2]. The current ql_syscall_futimesat() delegates to the utimensat path, so the second field is parsed as nanoseconds instead of microseconds. Please use a timeval-aware path like utimes() and multiply tv_usec by 1000 before calling os.utime(ns=...).

[libumem]

@xwings Finally got some free time to work on this again. Please take another look.

I've had to comment out a bunch of test cases (mostly arm64) that were having errors unrelated to anything I am touching in the codebase.