Update to optional NanoAOD version

[sofiagiappichini]

Every law/luigi command required --nanoAOD_version to be passed explicitly, even though the sample's nick is enough to determine unambiguously which version/directory a sample lives in. This work is related to the [sample database PR #62] (KIT-CMS/KingMaker_sample_database#62). This allows for multiple versions to be run from a single call.

What this PR does:

• processor/framework.py: Task.nanoAOD_version now defaults to "" instead of nanoAOD_v15. Adds two resolution helpers:
  • resolve_nanoAOD_version(relative_path, requested_version) — single-sample lookup by era/sample_type/nick.json, used by ConfigureDatasets.
  • resolve_sample_data(nick, requested_version) — per-sample lookup against each version's datasets.json, returns (version, sample_data), used by ProduceBase.
    Both return the passed-in version unchanged if one is given (--nanoAOD_version still works exactly as before — no directory searching happens in that case, and the old single-version error-if-not-found behavior is preserved). Otherwise they scan sample_database/nanoAOD_v*/ and auto-pick the version if exactly one contains the sample; if none or more than one do, they raise a clear error telling you to either fix the sample_database naming collision or pass --nanoAOD-version explicitly.
• processor/tasks/CROWNMain.py`: ConfigureDatasets.init resolves its own nanoAOD_version from era/sample_type/nick.
• processor/tasks/CROWNBase.py: ProduceBase no longer forces one shared nanoAOD_version for an entire sample_list. dataset_database is only pinned to a single version's datasets.json when --nanoAOD_version (or --dataset-database) is passed explicitly — matching the old behavior exactly. Otherwise, set_sample_data() resolves each requested sample's version individually, so a single sample_list can freely mix samples from different NanoAOD versions as long as each one is unambiguous on its own. The printed sample table also gained a "NanoAOD" column showing each sample's resolved version.
• processor/tasks/ProduceNtuples.py: requires() now passes each sample's own resolved nanoAOD_version explicitly into its CROWNFriend.req()/CROWNRun.req() call, so every downstream task in that sample's chain uses the correct version (and any further tasks it itself requires inherit that same concrete value automatically via law's .req() propagation).