Write the commit and timestamp into task_info.yaml

[rcannood]

Describe your changes

task_info.yaml was published as a verbatim copy of _viash.yaml:

def viash_file = meta.resources_dir.resolve("_viash.yaml")

so it could only ever contain things known at build time, and nothing in a published
result pinned it to the run that produced it. Two fields were missing:

• commit -- task_info.json has a field for it, but the closest _viash.yaml
  gets is version: build_main, which is the revision rather than a commit, and
  build/main moves.
• timestamp -- required by common/schemas/results_v4/task_info.json. Nothing
  was writing it, so render_results_report has been guessing: it parses a date out
  of a run_<date> results path, and falls back to the modification time of
  task_info.yaml if the path does not have one.

run_benchmark now reads _viash.yaml, adds both, and writes that out instead:

name: task_template
version: build_main
commit: 2dc5486b62cc1dc65b0abf371be898f8c246027e
timestamp: '2026-07-30T13:36:58Z'
...

workflow.commitId is null when Nextflow runs from a local checkout rather than a
revision it resolved itself, so that field is only added when we actually know it,
rather than writing a placeholder. The timestamp is the launch time rather than the
completion time, because workflow.complete is only known once the run is over, and
this closure runs before that -- it also lines up with the run_<date> directory name
that the reporting side was inferring from anyway.

Checked against Nextflow 26.04 that workflow.start is an OffsetDateTime, that
.toInstant().truncatedTo(SECONDS).toString() gives 2026-07-30T13:36:58Z, that
snakeyaml quotes it so it stays a string rather than a YAML timestamp, and that
yaml::read_yaml() reads it back as a character -- so it passes ajv's date-time
check. Comments in _viash.yaml do not survive the parse/serialise round trip, which
seemed a fair trade for a file that only ever gets parsed.

Same change as openproblems-bio/task_spatial_simulators#33 and
openproblems-bio/task_spatial_simulators#34, sent here as promised in the first of
those. The reporting side reads the commit field as of
openproblems-bio/openproblems#943; it already preferred the timestamp from
task_info.yaml over its --timestamp argument.

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!