Small refactoring for hermitian check

[blegat]

Less duplicate code is better

[codecov]

Codecov Report

Patch coverage: 100.00% and project coverage change: -0.01 ⚠️

Comparison is base (f4721e6) 98.12% compared to head (818c957) 98.12%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #3296      +/-   ##
==========================================
- Coverage   98.12%   98.12%   -0.01%     
==========================================
  Files          34       34              
  Lines        4740     4738       -2     
==========================================
- Hits         4651     4649       -2     
  Misses         89       89              

Impacted Files	Coverage Δ
src/sd.jl	100.00% <100.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[odow]

Less duplicate code is better

I don't agree. It moves the error into a utility function, when it should really be in build_variable. I think that makes it less readable. It's not obvious that _vectorize_complex_variables should throw an error if one of them is discrete.

[blegat]

If you have discrete constraints then in the vectorization itself, you should have to handle them, e.g., decide the value of the imag part of a boolean variable. Since the utility function does not handle that yet, it makes sense that it errors in that case.