Use ValueError for unsupported recursive glob in ADLS upload

[mitre88]

Part of the ongoing clean-up of broad AirflowException usages (enforced by the check-no-new-airflow-exceptions ratchet), following the pattern of #66279.

LocalFilesystemToADLSOperator.execute raised AirflowException when local_path contains an unsupported ** recursive glob — a plain input-validation error, now a ValueError. The known_airflow_exceptions.txt entry for the file drops from 1 to 0, and the existing test asserts the new type.

---

Was generative AI tooling used to co-author this PR?

• Yes — Claude Code (Fable 5)

Generated-by: Claude Code (Fable 5) following the guidelines

[potiuk]

Thanks — textbook version of this migration. An unsupported glob pattern is a caller-supplied argument error, which is exactly what ValueError is for, and the message is unchanged so nothing user-facing shifts except the type.

Three things done right that often get missed in this campaign: the now-unused AirflowException import is removed rather than left dangling; the known_airflow_exceptions.txt entry is deleted outright (count 1 -> 0) rather than decremented to a stale zero; and the test gains a match= argument, so it now pins the message as well as the type — strictly stronger than before.

Worth noting for completeness: ValueError isn't a subclass of AirflowException, so a Dag wrapping this operator in except AirflowException stops catching it. That's inherent to the migration rather than anything wrong here, and the ratchet file exists precisely to track it — just flagging it's a real behaviour change, not a pure refactor.

---

Drafted-by: Claude Code (Opus 5); reviewed by @potiuk before posting