Check that choices is non-empty in optimize_acqf_discrete

[Balandat]

Otherwise this produces confusing tensor dimension errors if no chocies are given.
Also applies ufmt style throughout.

[facebook-github-bot]

@Balandat has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[codecov]

Codecov Report

Merging #1228 (0e392dd) into main (9f1d62f) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##              main     #1228    +/-   ##
==========================================
  Coverage   100.00%   100.00%            
==========================================
  Files          120       122     +2     
  Lines        10439     10553   +114     
==========================================
+ Hits         10439     10553   +114     

Impacted Files	Coverage Δ
botorch/optim/optimize.py	100.00% <100.00%> (ø)
botorch/utils/containers.py	100.00% <0.00%> (ø)
botorch/test_functions/__init__.py	100.00% <0.00%> (ø)
botorch/models/utils/parse_training_data.py	100.00% <0.00%> (ø)
botorch/acquisition/multi_objective/__init__.py	100.00% <0.00%> (ø)
...h/test_functions/multi_objective_multi_fidelity.py	100.00% <0.00%> (ø)
...orch/acquisition/multi_objective/multi_fidelity.py	100.00% <0.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9f1d62f...0e392dd. Read the comment docs.