Dedup dataset metadata by dataset_id

[rcannood]

Describe your changes

The metadata channel wants one entry per dataset, but expresses that as "keep the log_cp10k ones":

  meta_ch = dataset_ch
    // only keep one of the normalization methods
    | filter{ id, state ->
      state.rna_norm == "log_cp10k"
    }

That holds only while log_cp10k is among the normalizations we materialise. The day a dataset arrives under a different one -- or log_cp10k gets renamed -- the filter passes nothing, joinStates never fires, and output_dataset_info / output_method_configs / output_metric_configs / output_task_info are simply absent from the final state. Silently, since an empty channel isn't an error.

De-duplicating on dataset_id instead says the intended thing directly and doesn't care how many normalizations there are. Checked the Groovy under nextflow:

kept 2 of 3: [[dataset_id:a, x:1], [dataset_id:b, x:2]]

(three states, two datasets, normalization_id stripped as before.)

No behavioural change today, since log_cp10k is the only normalization in play.

Part of a series of PRs coming out of a pre-run review of the benchmark.

Checklist before requesting a review

• I have performed a self-review of my code

• Check the correct box. Does this PR contain:

  • Breaking changes
  • New functionality
  • Major changes
  • Minor changes
  • Bug fixes

• Proposed changes are described in the CHANGELOG.md

• CI Tests succeed and look good!