Replace CASTEP_MD parser with castep_outputs

[oerc0122]

Description of work
Replace custom parser with library designed for purpose.

Fixes
N/A

To test
All conversion tests should work as normal.

[MBartkowiakSTFC]

One observation so far is that this implementation of the converter is more strict than the original MDANSE CASTEP converter. The conversion will fail if the header is present, but the empty line after the END header is missing.

2026-05-06 10:23:38,369 - CRITICAL - process[24450] - IJob 541 - Job failed with traceback: Traceback (most recent call last):
IJob._runner[mode](self, prog=prog_bar)
File "MDANSE/Src/MDANSE/Framework/Jobs/IJob.py", line 389, in _run_singlecore
idx, result = self.run_step(index)
^^^^^^^^^^^^^^^^^^^^
File "MDANSE/Src/MDANSE/Framework/Converters/CASTEP.py", line 111, in run_step
frame = next(self.frames)
^^^^^^^^^^^^^^^^^
File "MDANSE/Src/MDANSE/Framework/Parsers/CASTEP_MD.py", line 58, in frames
yield from map(self.parse_frame, parser)
File "MDANSE/Src/MDANSE/Framework/Parsers/CASTEP_MD.py", line 63, in parse_frame
accum["time"] = frame["time"] * AUT
~~~~~~~~~~~~~~^~~~~
TypeError: can't multiply sequence by non-int of type 'float'

[oerc0122]

I have a PR in-place in castep_outputs (oerc0122/castep_outputs#264) to fix that, but yes for now it strictly follows the definition.

[MBartkowiakSTFC]

I installed castep_outputs==0.3.3, but on my machine the conversion still fails with the same error posted above, both for input files without the header and those that have the header but don't have an empty line afterwards.

[oerc0122]

This is now blocked by oerc0122/castep_outputs#270