Min max scaler util

[ebamberg]

Description

This is MinMaxScaler implementation similar to sklearn.preprocessing.MinMaxScaler.
wanted to share this with you, just in-case that such tooling is from interest to add to the djl repo.

I needed such function in some time-series-forecast-demo and I haven't seen a similar implementation in the repo. So mainly matches functionality I needed.

• If this change is a backward incompatible change, why must this change be made?
• Interesting edge cases to note here

[codecov-commenter]

Codecov Report

Merging #986 (c13e322) into master (f2e56d5) will decrease coverage by 0.00%.
The diff coverage is 72.41%.

@@             Coverage Diff              @@
##             master     #986      +/-   ##
============================================
- Coverage     70.04%   70.03%   -0.01%     
- Complexity     5085     5105      +20     
============================================
  Files           502      503       +1     
  Lines         22509    22579      +70     
  Branches       2355     2366      +11     
============================================
+ Hits          15766    15813      +47     
- Misses         5489     5499      +10     
- Partials       1254     1267      +13     

Impacted Files	Coverage Δ
...c/main/java/ai/djl/training/util/MinMaxScaler.java	72.41% <72.41%> (ø)
.../src/main/java/ai/djl/pytorch/engine/PtEngine.java	47.61% <0.00%> (-8.64%)	⬇️
api/src/main/java/ai/djl/BaseModel.java	73.64% <0.00%> (-1.01%)	⬇️
api/src/main/java/ai/djl/inference/Predictor.java	90.38% <0.00%> (+2.88%)	⬆️

---

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 f2e56d5...c13e322. Read the comment docs.

[ebamberg]

added a function to detach to MinMaxScaler from NDManager