Fix regression preprocessing

[YamLyubov]

The problem was that tuner was returning already fitted pipeline. And when final pipeline (after composition and tuning) was fitted on whole data, operations in nodes was not refitted. This was causing the error.

Now tuner returns unfitted pipeline.

Fixes #942

[codecov]

Codecov Report

Merging #955 (88e8a09) into master (28216cc) will increase coverage by 0.19%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #955      +/-   ##
==========================================
+ Coverage   88.22%   88.41%   +0.19%     
==========================================
  Files         205      205              
  Lines       13409    13312      -97     
==========================================
- Hits        11830    11770      -60     
+ Misses       1579     1542      -37     

Impacted Files	Coverage Δ
fedot/core/pipelines/tuning/tuner_builder.py	97.10% <ø> (ø)
...t/core/optimisers/objective/data_objective_eval.py	92.20% <100.00%> (-0.20%)	⬇️
fedot/core/pipelines/tuning/tuner_interface.py	91.07% <100.00%> (ø)
fedot/explainability/explainer_template.py	75.00% <0.00%> (-5.00%)	⬇️
...ot/core/composer/gp_composer/specific_operators.py	88.13% <0.00%> (-3.39%)	⬇️
...on_implementations/models/discriminant_analysis.py	92.45% <0.00%> (-2.00%)	⬇️
...implementations/data_operations/sklearn_filters.py	94.49% <0.00%> (-1.84%)	⬇️
fedot/core/repository/tasks.py	72.72% <0.00%> (-1.56%)	⬇️
fedot/core/caching/base_cache.py	66.66% <0.00%> (-1.52%)	⬇️
fedot/core/data/multi_modal.py	84.37% <0.00%> (-1.21%)	⬇️
... and 50 more

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

[nicl-nno]

В какой-то из тестов бы ещё добавить проверку что пайплайн не-fitted после тюнинга.