feat: parameter customization for general inference

[alexander-held]

This is an extension of #320, adding init_pars / fix_pars / par_bounds also to the remaining fit API: fit.ranking,
fit.scan, fit.limit, fit.significance.

Small improvement added: set lower POI bound for limit to 0.0 only if existing bound is smaller than 0.

* added parameter customization options to remaining fit API
* supported keyword arguments are the same as for fit.fit: init_pars, fix_pars, and par_bounds
* set lower POI bound for limit to 0 only if default is below zero

[codecov]

Codecov Report

Merging #321 (a00b0bf) into master (76c77e4) will not change coverage.
The diff coverage is 100.00%.

@@            Coverage Diff            @@
##            master      #321   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files           23        23           
  Lines         1880      1881    +1     
  Branches       304       305    +1     
=========================================
+ Hits          1880      1881    +1     

Impacted Files	Coverage Δ
src/cabinetry/fit/__init__.py	100.00% <100.00%> (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 76c77e4...a00b0bf. Read the comment docs.