feat: grad_scale option for fit

[jiangyi15]

$\min [ L(x) ] => \min [ k L(x)] $, k is a scale factor.
It could improve fit stability when the $|k L(x)|$ is not too large or small.

config.fit(grad_scale=1e-3)

A test of scale vs fit cost time and number of function evaluations. The final $- \ln L \approx -5187$.

[codecov]

Codecov Report

Merging #97 (05fa8ae) into dev (b8e892c) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##              dev      #97   +/-   ##
=======================================
  Coverage   75.04%   75.05%           
=======================================
  Files         106      106           
  Lines       15270    15271    +1     
  Branches     2844     2844           
=======================================
+ Hits        11460    11461    +1     
  Misses       3122     3122           
  Partials      688      688           

Flag	Coverage Δ
unittests	75.05% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files Changed	Coverage Δ
tf_pwa/config_loader/config_loader.py	66.82% <ø> (ø)
tf_pwa/fit.py	22.75% <100.00%> (ø)
tf_pwa/fit_improve.py	56.43% <100.00%> (+0.12%)	⬆️