ReadVarsESO silently fails with wrong output names

[airallergy]

From discussion with @shyamamrith.

When a wrong output name is specified, ReadVarsESO will indicate that the output is not found in readvars.audit, but still run successfully with exit code 0, and create a csv file with a single line.

This will lead to an error in the method

sober/sober/_io_managers.py

Lines 546 to 548 in a2548e9

	def _record_final(
	self, level: AnyCoreLevel, record_dir: Path, uids: AnyUIDs
	) -> None:

where next() in

sober/sober/_io_managers.py

Line 614 in a2548e9

record_rows[i] += next(reader)[1:]

will fail because of reaching the end of iterator.

Unfortunately this is not caught due to multiprocessing plus hidden by the with open() context. But rather, this causes scheduled in

sober/sober/_io_managers.py

Line 669 in a2548e9

for (job_uid, _), _ in zip(batch, scheduled, strict=True):

to be empty, hence failing the strict zip, which is completely uninformative.

Consider how to provide more informative error message on this.