test: cover TX_TIME and TIME/ASC sort paths on v0 listing

[odesenfans]

Summary

Adds two black-box tests against get_matching_posts_legacy to cover sort paths that previously had no exercise:

• test_get_matching_posts_legacy_sort_order_time: asserts both ASC and DESC orderings on the default SortBy.TIME sort.
• test_get_matching_posts_legacy_sort_order_tx_time: seeds two chain_txs + message_confirmations rows with controlled timestamps and a third unconfirmed post, then asserts ASC/DESC orderings including the nullsfirst/nullslast branches.

Coverage on aleph.db.accessors.posts from tests/db/test_posts.py alone: 87% → 91%. The entire if sort_order: block in filter_post_select_stmt is now covered.

Branched directly off main (not on top of #1133) so the tests can be verified against the current code first; #1133 will rebase on this once merged.

Test plan

• hatch run testing:test tests/db/test_posts.py -v (11 tests pass against current main)

🤖 Generated with Claude Code

[foxpatch-aleph]

Clean PR that adds two black-box tests exercising the TIME and TX_TIME sort paths on the v0 listing endpoint. The tests are correctly structured, follow existing patterns in the file, and cover all branches of the sort logic in filter_post_select_stmt including the nullsfirst/nullslast paths for TX_TIME. No logic errors or security concerns found.

tests/db/test_posts.py (line 515): Minor: sort_by is not specified here, so it falls through to the else branch in filter_post_select_stmt which treats it as TIME sort. Consider passing sort_by=SortBy.TIME explicitly for clarity, or leave as-is since it tests the implicit default — either is acceptable.

[foxpatch-aleph]

Clean PR that adds two well-structured integration tests covering previously untested sort paths (TIME and TX_TIME with ASC/DESC) in get_matching_posts_legacy. The tests are correct, follow existing patterns, and properly exercise the nullsfirst/nullslast branches for TX_TIME sorting. No bugs, security issues, or logic errors found.