Change NDArray.toString() output

[frankfliu]

Change-Id: I4c0b29e56fd6ed214656cf2681337e595ecda4e3

Description

Brief description of what this PR is about

• 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 #1142 (04b891f) into master (e0aaa4b) will increase coverage by 0.04%.
The diff coverage is 60.60%.

@@             Coverage Diff              @@
##             master    #1142      +/-   ##
============================================
+ Coverage     70.12%   70.16%   +0.04%     
- Complexity     5329     5332       +3     
============================================
  Files           513      513              
  Lines         23696    23687       -9     
  Branches       2545     2546       +1     
============================================
+ Hits          16617    16621       +4     
+ Misses         5722     5707      -15     
- Partials       1357     1359       +2     

Impacted Files	Coverage Δ
...ne/src/main/java/ai/djl/dlr/engine/DlrNDArray.java	35.29% <0.00%> (+3.71%)	⬆️
...st/src/main/java/ai/djl/ml/xgboost/XgbNDArray.java	59.61% <0.00%> (+4.25%)	⬆️
...ain/java/ai/djl/onnxruntime/engine/OrtNDArray.java	36.58% <0.00%> (+2.49%)	⬆️
...ain/java/ai/djl/paddlepaddle/engine/PpNDArray.java	50.00% <0.00%> (+4.76%)	⬆️
.../main/java/ai/djl/tflite/engine/TfLiteNDArray.java	36.61% <0.00%> (+2.40%)	⬆️
...rc/main/java/ai/djl/ndarray/internal/NDFormat.java	81.62% <63.63%> (-2.29%)	⬇️
api/src/main/java/ai/djl/ndarray/NDArray.java	75.72% <100.00%> (+0.14%)	⬆️
...e/src/main/java/ai/djl/mxnet/engine/MxNDArray.java	86.72% <100.00%> (ø)
...src/main/java/ai/djl/pytorch/engine/PtNDArray.java	80.69% <100.00%> (-0.05%)	⬇️
.../main/java/ai/djl/tensorflow/engine/TfNDArray.java	84.34% <100.00%> (ø)
... and 2 more

---

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 e0aaa4b...04b891f. Read the comment docs.

[zachgk]

This makes the default maxSize 0, which means it won't show any NDArrays. We should probably have a larger default to allow some to be displayed. Maybe 100 like MxNDarray before?

[frankfliu]

Currently, toString() is called by default in debugger, retrieving the data makes the debugger very slow. The default should be 0. User can manually configure debugger to use toDebugString() when then expend the data view, this will be on-demand and won't impact performance.

We display the content mostly in Jupyter notebook, I'm consider add the property in the IJava kernel to workaround this issue.

[zachgk]

It would also affect println statements as well. Here is another option. What if we added something like this check to see if the JVM is running in debug mode?

[frankfliu]

Now we won't print ndarray content in debugger