Add support for mixed precision training

[NihalHarish]

Description of changes:

• Add support for Autocast variables that are created when mixed precision training is enabled.

Style and formatting:

I have run pre-commit install to ensure that auto-formatting happens with every commit.

Issue number, if available

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

[codecov-io]

Codecov Report

Merging #378 into master will decrease coverage by 1.53%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #378      +/-   ##
==========================================
- Coverage   87.06%   85.52%   -1.54%     
==========================================
  Files          86       86              
  Lines        6507     6514       +7     
==========================================
- Hits         5665     5571      -94     
- Misses        842      943     +101     

Impacted Files	Coverage Δ
smdebug/tensorflow/keras.py	92.32% <100.00%> (ø)
smdebug/tensorflow/tensor_ref.py	88.70% <100.00%> (ø)
smdebug/tensorflow/utils.py	88.32% <100.00%> (+0.43%)	:arrow_up:
smdebug/mxnet/graph.py	25.80% <0.00%> (-64.52%)	:arrow_down:
smdebug/core/access_layer/base.py	64.28% <0.00%> (-21.43%)	:arrow_down:
smdebug/core/modes.py	55.00% <0.00%> (-15.00%)	:arrow_down:
smdebug/core/access_layer/utils.py	77.27% <0.00%> (-13.64%)	:arrow_down:
smdebug/mxnet/collection.py	76.66% <0.00%> (-13.34%)	:arrow_down:
smdebug/core/logger.py	66.12% <0.00%> (-12.91%)	:arrow_down:
smdebug/core/tfrecord/record_reader.py	78.00% <0.00%> (-12.00%)	:arrow_down:
... and 14 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 7470da7...da8d716. Read the comment docs.