[api] Fixes memory leak in hybrid engine

[frankfliu]

Change-Id: I63f76479ea5acc55cd7725cedf13563157cac6d4

Description

Fixes #1516

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

[lanking520]

Could you please elaborate more where did you find the memory leaks?

[codecov-commenter]

Codecov Report

Merging #1518 (3e6b94d) into master (bb5073f) will decrease coverage by 1.27%.
The diff coverage is 50.76%.

@@             Coverage Diff              @@
##             master    #1518      +/-   ##
============================================
- Coverage     72.08%   70.80%   -1.28%     
- Complexity     5126     5326     +200     
============================================
  Files           473      493      +20     
  Lines         21970    23275    +1305     
  Branches       2351     2548     +197     
============================================
+ Hits          15838    16481     +643     
- Misses         4925     5514     +589     
- Partials       1207     1280      +73     

Impacted Files	Coverage Δ
api/src/main/java/ai/djl/modality/cv/Image.java	69.23% <ø> (-4.11%)	⬇️
...i/djl/modality/cv/translator/BigGANTranslator.java	21.42% <ø> (-5.24%)	⬇️
...odality/cv/translator/BigGANTranslatorFactory.java	33.33% <0.00%> (+8.33%)	⬆️
...nslator/InstanceSegmentationTranslatorFactory.java	14.28% <0.00%> (-3.90%)	⬇️
.../modality/cv/translator/YoloTranslatorFactory.java	8.33% <0.00%> (-1.67%)	⬇️
...i/djl/modality/cv/translator/YoloV5Translator.java	5.69% <0.00%> (ø)
...odality/cv/translator/YoloV5TranslatorFactory.java	8.33% <0.00%> (-1.67%)	⬇️
...pi/src/main/java/ai/djl/ndarray/BytesSupplier.java	54.54% <0.00%> (-12.13%)	⬇️
...l/training/loss/SigmoidBinaryCrossEntropyLoss.java	64.00% <0.00%> (ø)
api/src/main/java/ai/djl/translate/Translator.java	50.00% <0.00%> (-50.00%)	⬇️
... and 183 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 2a1441c...3e6b94d. Read the comment docs.

[nguyenvietyen]

Could you please elaborate more where did you find the memory leaks?

My colleague @andreabrduque discovered it when we performed a load-test (as part of readying for production) of a new model. It causes the nodes to go OOM. She then distilled it to a ticket: #1516

[frankfliu]

@nguyenvietyen
I added root cause in the ticket