fix: [libpthreadglue] tighten error handling in OS-abstraction layer

[fjtrujy]

Summary

• pte_osThreadCreate used to malloc the stack and call CreateThread, then NULL-check the stack and check threadId < 0 after __threadInfo[threadId] was already written — a negative id caused an OOB write. The check for the cancelSem semaphore was also missing, and the failure paths leaked cancelSem and the stack.
• Reorder so each resource is validated immediately after it is allocated and cleaned up in reverse order on failure.
• Propagate CreateSema failures from pte_osMutexCreate and pte_osSemaphoreCreate as PTE_OS_NO_RESOURCES instead of returning a negative handle through *pHandle as if it were a valid semaphore.

Related

• Companion pthread-embedded PR: Reject NULL / uninitialised mutex handles with EINVAL pthread-embedded#2 — adds the NULL-handle rejection and mutex9 test referenced in the test plan.

Test plan

• Build ee/libpthreadglue cleanly.
• Run pthread-embedded test suite on PS2 — exercises thread/mutex/sema creation paths and now also the NULL-handle paths from the companion pthread-embedded PR.
• Smoke-test a sample that creates pthreads (e.g. a port that already uses libpthreadglue).

[uyjulian]

Lgtm