fix: preserve TextMentionTermination.sources across serialization round-trip

[Humphrey (HumphreySun98)]

Why are these changes needed?

TextMentionTermination accepts a sources argument that restricts which
message sources are checked for the termination text:

term = TextMentionTermination("stop", sources=["assistant"])

However TextMentionTerminationConfig only stored text, and
_to_config/_from_config only serialized/restored text. So a condition
configured with sources silently loses that restriction after a declarative
dump_component() / load_component() round-trip, and the reloaded condition
then fires on any source (including user or other agents) — a behavior
change on save/load.

The sibling SourceMatchTermination already round-trips its sources field
correctly (and the existing declarative test asserts it), which makes this an
inconsistency rather than an intentional omission.

This PR adds sources to TextMentionTerminationConfig and serializes/restores
it in _to_config/_from_config, with a regression test covering both the
sources-restricted and unrestricted cases.

Related issue number

N/A — found during review. Fixes a serialization round-trip data loss;
SourceMatchTermination is the correct reference behavior.

Related PRs

• fix: preserve TextMentionTermination sources in config #7837 (2026-06-15) predates this PR and makes the equivalent change (adds sources to the config and round-trips it). I did not find it before opening this one. Differences are minor: fix: preserve TextMentionTermination sources in config #7837 normalizes sources to a list in __init__, this PR does so in _to_config, and this PR's test also covers the unrestricted (sources=None) case. Either fixes the bug; happy for maintainers to merge whichever and close the other.
• fix: TextMentionTermination drops sources in component serialization; docstring and typo fixes #8088 (2026-08-23) bundles the same fix with unrelated changes across several other files.

Checks

• I've included any doc changes needed for https://microsoft.github.io/autogen/ (none required for this change).
• I've added tests (if relevant) corresponding to the changes introduced in this PR.
• I've made sure all auto checks have passed (ran ruff, mypy, and the relevant pytest locally).

[ErenAta16 (ErenAta16)]

Ran the round-trip test — passes for both the sources-restricted and unrestricted cases. Worth noting this brings TextMentionTermination in line with SourceMatchTermination, which the PR description says already round-trips sources correctly — so this was an inconsistency between two conceptually similar termination conditions rather than a one-off oversight, which matches what I'd expect from how these config classes tend to drift when a field gets added to the runtime class but not mirrored into _to_config/_from_config.

[Humphrey (HumphreySun98)]

Closing and immediately reopening to re-create the expired CI workflow runs (explanation follows).

[Humphrey (HumphreySun98)]

Explanation for the reopen above: the workflow runs created when this PR opened in July had expired (GitHub expires unapproved fork runs after 30 days) and been purged, so the PR carried zero runs. There was therefore nothing for a maintainer to approve, and the 18 required status checks could never have reported.

Reopening at the same commit re-created them, with no commits changed and nothing force-pushed. This PR now has 7 runs in action_required, so "Approve and run workflows" is actionable. Full write-up in #7931.