check-setup: validate input datasets at their configured path_in_babs

[yibeichan]

Closes #325.

The bug

babs_check_setup probes a hardcoded <analysis>/inputs/data before validating
the input datasets:

temp_list = get_immediate_subdirectories(op.join(self.analysis_path, 'inputs/data'))

babs init clones each input dataset to its configured path_in_babs, so on any
project not using the default layout this raises a raw FileNotFoundError out of
os.listdir — not even the friendly ValueError on the line below, which only
fires for a directory that exists but is empty.

Since #369 this affects a layout BABS itself generates. The BIDS-study layout
puts inputs under sourcedata/<name>, so babs init writes:

input_datasets:
  BIDS:
    path_in_babs: 'sourcedata/BIDS'
  NIDM:
    path_in_babs: 'sourcedata/NIDM'

and the resulting analysis directory has sourcedata/{BIDS,NIDM} and no inputs/
at all. babs check-setup cannot run on such a project, which means losing the
test-job and environment checks that do work — we currently skip check-setup
entirely on every run.

The fix

The pre-check was already redundant. The per-dataset loop immediately below
resolves in_ds.babs_project_analysis_path and raises for a missing directory
or a non-DataLad dataset, using the configured path — so it covers everything
the hardcoded probe did, correctly.

The one condition the probe uniquely covered is "no input datasets configured",
which is a property of the config rather than of the filesystem. This PR checks
that directly and drops the path probe, so there is no second path-resolution
to keep in sync with path_in_babs.

get_immediate_subdirectories was imported here only for that call. It is now
unused within babs/, but its definition in utils.py is left in place — it is
public API and tests/test_utils.py::test_get_immediate_subdirectories covers it.

Verification

Against a real BIDS-study-layout project (path_in_babs: sourcedata/BIDS and
sourcedata/NIDM), on ea5d6f0:

OLD pre-check: FileNotFoundError -> .../analysis/inputs/data
NEW pre-check: passes, 2 input dataset(s)
   BIDS   path_in_babs=sourcedata/BIDS   exists=True is_datalad=True
   NIDM   path_in_babs=sourcedata/NIDM   exists=True is_datalad=True

ruff check and ruff format --check are clean on the changed file.

On tests: I did not add one. The existing tests/test_check_setup.py cases are
integration tests built on the babs_project_* fixtures and the simbids apptainer
image, which I can't run in my environment, and there is no study-layout fixture to
extend. Happy to add a regression test if you can point me at the fixture you'd want
it built on — a path_in_babs-varying project fixture would cover this directly.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 0% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 79.38%. Comparing base (ea5d6f0) to head (f685e8f).

Files with missing lines	Patch %	Lines
babs/check_setup.py	0.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #407      +/-   ##
==========================================
- Coverage   79.39%   79.38%   -0.01%     
==========================================
  Files          17       17              
  Lines        2169     2168       -1     
  Branches      384      384              
==========================================
- Hits         1722     1721       -1     
  Misses        302      302              
  Partials      145      145              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.