Fix cp handling for deleting destination entries

[mother-6000]

Closes #

Please check if the PR fulfills these requirements

• Tests for the changes have been added (for bug fixes / features)
• Docs have been added / updated (for bug fixes / features)
• CHANGELOG.md has been updated (for bug fixes / features / docs)

What kind of change does this PR introduce?

Bug fix.

What was changed?

• pre-check destination bucket entries before bucket-to-bucket cp starts and mark entries already in Deleting state as failed instead of trying to write into them
• keep --from-last entry start overrides based on a single destination entry listing pass
• add a race-safe fallback for late 409 Conflict responses by re-checking destination entry status and failing the entry when it has transitioned to Deleting
• bump the CLI version to 0.12.2 and document the fix in CHANGELOG.md

Related issues

• user-reported bucket-to-bucket cp against deleting destination entries silently counting all records as skipped after backend 409 Conflict responses

Does this PR introduce a breaking change?

No.

Other information:

Validation performed:

• cargo test --quiet --no-run
• cargo test --quiet test_build_blocked_entry_errors_marks_deleting_entries
• cargo test --quiet test_entry_tree_shows_failure_for_entries_without_copied_records
• cargo test --quiet test_message_shows_skipped_only_when_non_zero

Note: the broader cp test suite in this environment depends on the local ReductStore test server and was not runnable here without that service.