[PWGHF] HFInvMassFitter: code cleaning and refactoring

[lubynets]

Bugfix:

• Before the json configurable UseLikelihood did not have any effect: independently on its value true or false, always likelihood fit was performed. This bug is fixed now. Note: it seems that in Chi2 mode (when RooAbsPdf::chi2FitTo() function is called) setting the fitting range has no effect. This problem is to be investigated, any help will be highly appreciated;

Code cleaning and refactoring:

• Reduced code repetition in several places;
• Used enums instead of magic values for fit functions and config parameters;
• Used std::array instead of std::vectors when the size is known at the compilation time;
• ROOT types replaced with standard c++ types (with an exception for TString in several places);
• Got rid of compilation warnings;
• Deleted the default constructor;
• Factored out into a function:
  • open file and check for nullptr;
  • get object from file and check for nullptr;
  • read variables and vectors from json configurables; avoid crash when the field is optional and crash with explicit error message when the field is mandatory;
  • check whether the vector of configurables has the same size as number of fitting ranges (or zero if the vector is optional);
• Got rid of unnecessary histograms cloning when passing them from file to the fitter and when rebinning them;
• Function implementations, which occupy more than 1 line, moved from header to the source file.

[github-actions]

O2 linter results: ❌ 1 errors, ⚠️ 0 warnings, 🔕 0 disabled

[vkucera]

Hi @lubynets , thanks for the extensive improvement. Can you please also fix the issues reported in the bug tracker?