Chem: add #meta.role: transform to stand-alone scripts for datasync support

[Aleksashka11]

Drafted by Grokky from a Slack thread.

Root cause: File: packages/Chem/scripts/stand-alone/butina_cluster.py (and sibling scripts murcko_scaffold.py, filter_catalogs.py)

What breaks: When a project is saved with datasync enabled and then reopened, the Butina-cluster step is not replayed — the platform sees only a stale, frozen result column instead of re-running the function.

Where: The datasync subsystem decides whether a function result can be replayed by checking for the #meta.role: transform script header. Every biochemical-calculator script that correctly supports datasync carries this tag (calc_logP.py, calc_logS.py, calc_pI.py, calc_pKas.py). All three stand-alone scripts (butina_cluster.py, murcko_scaffold.py, filter_catalogs.py) are missing it, even though they also use {action:join(data)} in their #output directive (i.e. they produce joined result columns exactly like the calculators do).

Why: butina_cluster.py was added to the top-menu (Chem | Analyze | Butina Cluster...) without the matching datasync annotation. Without #meta.role: transform, the platform cannot identify the function as re-runnable during project restore, so it displays the state frozen at save-time rather than recalculating — producing "incorrect results for Butina cluster" as reported.

Note: calc_logD.py has the same annotation gap, but that is outside the scope of GROK-20508.

Reproduced on dev: Reproduction requires save/restore of a project with datasync through the full UI, which is not automatable through the JS API alone. Root cause is established purely by static code comparison.

Thread: https://datagrok.slack.com/archives/C04BF3YM6CF/p1785227533092579?thread_ts=1785227533.092579&cid=C04BF3YM6CF

Auto-drafted fix — review and test before merging.