Support _estimator_type.

[trivialfis]

For more info, see: https://scikit-learn.org/stable/developers/develop.html#estimator-types

• Model trained from dask can be loaded by single node skl interface. Close [FEATURE] convert dask -> non-dask model #6547 .

[codecov-io]

Codecov Report

Merging #6582 (b948de5) into master (b2246ae) will increase coverage by 0.11%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #6582      +/-   ##
==========================================
+ Coverage   80.10%   80.21%   +0.11%     
==========================================
  Files          13       13              
  Lines        3604     3609       +5     
==========================================
+ Hits         2887     2895       +8     
+ Misses        717      714       -3     

Impacted Files	Coverage Δ
python-package/xgboost/sklearn.py	89.34% <100.00%> (+0.33%)	⬆️
python-package/xgboost/tracker.py	93.98% <0.00%> (-1.13%)	⬇️
python-package/xgboost/dask.py	81.38% <0.00%> (+0.84%)	⬆️

---

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 b2246ae...b948de5. Read the comment docs.