refactor: extract telemetry span lifecycle helpers

[deepujain]

• I, the PR author, have personally reviewed every line of this PR.

What does this PR do?

Large training modules are harder to review and maintain when telemetry lifecycle code is mixed into the training loop. This change moves OpenTelemetry span state and lifecycle helpers from megatron/training/training.py into megatron/training/telemetry_spans.py, while leaving the training call sites unchanged.

The extracted module owns startup and checkpoint-interval spans, shutdown state, and one-time atexit and SIGTERM hook installation. Startup timestamps are passed explicitly to avoid a circular dependency on training.py.

CPU-only tests cover startup closure, interval rerooting, one-time hook installation, hard SIGTERM teardown, and graceful SIGTERM flushing.

Issue tracking

Fixes #6631

Contribution process

Pre-checks

• I have added relevant unit tests
• I have added relevant functional tests
• I have added proper typing to my code Typing guidelines
• I have added relevant documentation
• I have run the autoformatter.sh on my PR

Validation

• Five focused lifecycle regressions passed in the earlier isolated CPU harness; the extracted module and tests are byte-identical on the repaired head.
• python3 -m py_compile passed for all changed Python files on the repaired head.
• git diff --check passed.
• Independent exact-head review found no blocking lifecycle, compatibility, or conflict-resolution defects.

The repaired commit is rebased on current main and signed. Full package test collection on the local macOS host is blocked by the repository's unavailable Triton/GPU dependencies, so hosted GPU validation remains a maintainer-run gate.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.