Get FC import from CASTEP 25.1 working

[ajjackson]

Closes #294

• CASTEP 25.1 adds extra fields to BORN and DIELECTRIC sections of castep_bin file; read (and discard) this information
• It is also now possible to create a BORN section without corresponding DIELECTRIC, using a finite-displacement scheme with ultrasoft pseudopotentials. That throws off some Euphonic logic which assumes you would only ever have BORN and DIELECTRIC data at the same time.
• As FC behaviour isn't really defined when you only have one, let's discard it in that case.

I need to test a wider range of import/present/not-present cases to be sure this will always do the right thing.

The good news is that most cases seem to work without this update; the main issue is with Berry-phase Born charge calculations.

[github-actions]

Test Results

   21 files  +  3     21 suites  +3   1h 3m 43s ⏱️ + 5m 0s
1 083 tests +  2  1 077 ✅ +  2   6 💤 ±0  0 ❌ ±0 
9 709 runs   - 765  9 651 ✅  - 765  58 💤 ±0  0 ❌ ±0 

Results for commit 7f7bab3. ± Comparison against base commit 1d9874a.

♻️ This comment has been updated with latest results.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.04%. Comparing base (48f8d11) to head (799b1d1).
Report is 1 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #364   +/-   ##
=======================================
  Coverage   96.03%   96.04%           
=======================================
  Files          31       31           
  Lines        4214     4219    +5     
  Branches      643      645    +2     
=======================================
+ Hits         4047     4052    +5     
  Misses         95       95           
  Partials       72       72           

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

[ajjackson]

Note that in the new test reference data:

• For a CASTEP 25 calculation which loads from an external BORN file, the Born effective charges and dielectric tensor are included in the JSON as "usual"
• For a CASTEP 25 calculation which calculations Born effective charges by finite displacements and does not have a corresponding dielectric tensor, the Born effective charges are omitted from the Force Constants. (They cannot be used to correct frequencies and no NAC term has been subtracted from the Force Constants, so it would be misleading to include them.)

[ajjackson]

I deliberately pushed a failing commit with updated tests, to verify that the new tests cover a real problem. The actual fix is in the following commit.

[ajjackson]

Looks like I got slightly carried away with whitespace

[oerc0122]

Looks sensible and simple. Concerns about changing the process generally when it should presumably only affect 25.1 files?

[ajjackson]

Forgot the changelog 😅

[oerc0122]

And here I thought FC stood for Forget the Changelog.

Looks good!