Cache the output of is_framework_version_supported

[NihalHarish]

Description of changes:

• The value of is_framework_version_supported needs to be computed only once since version checks are expensive and the value of the framework during execution will not change.

Testing

I have validated that this change reduces the perf impact of the function using pyinstrument. See code below:

from torch.utils.smdebug import get_smdebug_hook

NUM_ITER = 10000

for i in range(NUM_ITER):
    hook = get_smdebug_hook()

Without change:

Program: test_smdebug_hook.py

1.300 <module>  <string>:1
   [4 frames hidden]  <string>, runpy
      1.300 _run_code  runpy.py:64
      └─ 1.300 <module>  test_smdebug_hook.py:1
         ├─ 0.782 get_smdebug_hook  torch/utils/smdebug.py:28
         │  ├─ 0.351 error_handling_agent  smdebug/core/error_handling_agent.py:99
         │  │  └─ 0.349 get_hook  smdebug/pytorch/singleton_utils.py:17
         │  │     ├─ 0.248 validate_training_job  smdebug/core/config_validator.py:65
         │  │     │  └─ 0.242 _validate_training_environment  smdebug/core/config_validator.py:25
         │  │     │     └─ 0.240 is_framework_version_supported  smdebug/core/utils.py:641
         │  │     │        └─ 0.238 is_current_version_supported  smdebug/pytorch/utils.py:91
         │  │     │           └─ 0.223 parse  packaging/version.py:49
         │  │     │                 [24 frames hidden]  packaging, <built-in>, <string>

With Change:

Program: test_smdebug_hook.py

1.021 <module>  <string>:1
   [4 frames hidden]  <string>, runpy
      1.021 _run_code  runpy.py:64
      └─ 1.021 <module>  test_smdebug_hook.py:1
         ├─ 0.519 get_smdebug_hook  torch/utils/smdebug.py:28
         │  ├─ 0.261 <module>  smdebug/__init__.py:2
         │  │  └─ 0.261 <module>  smdebug/core/collection.py:2
         │  │     └─ 0.261 <module>  smdebug/core/reduction_config.py:2
         │  │        └─ 0.260 <module>  smdebug/core/utils.py:2
         │  │           ├─ 0.193 <module>  smdistributed/modelparallel/torch/__init__.py:2
         │  │           │     [655 frames hidden]  smdistributed, smexperiments, pkg_res...
         │  │           ├─ 0.041 <module>  requests/__init__.py:8
         │  │           │     [122 frames hidden]  requests, urllib3, re, sre_compile, s...
         │  │           └─ 0.021 <module>  horovod/torch/__init__.py:17
         │  │                 [64 frames hidden]  horovod, psutil, <built-in>, collecti...

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-commenter]

Codecov Report

Merging #508 (42117b0) into master (314c091) will decrease coverage by 1.73%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #508      +/-   ##
==========================================
- Coverage   75.89%   74.16%   -1.74%     
==========================================
  Files         126      116      -10     
  Lines       10872    10527     -345     
==========================================
- Hits         8251     7807     -444     
- Misses       2621     2720      +99     

Impacted Files	Coverage Δ
smdebug/core/utils.py	80.17% <100.00%> (-1.12%)	:arrow_down:
smdebug/rules/action/message_action.py	81.92% <0.00%> (-15.67%)	:arrow_down:
smdebug/xgboost/utils.py	0.00% <0.00%> (-14.76%)	:arrow_down:
smdebug/profiler/tf_profiler_parser.py	54.54% <0.00%> (-11.58%)	:arrow_down:
smdebug/rules/action/stop_training_action.py	54.68% <0.00%> (-7.82%)	:arrow_down:
smdebug/mxnet/collection.py	73.33% <0.00%> (-6.67%)	:arrow_down:
smdebug/mxnet/utils.py	59.37% <0.00%> (-6.25%)	:arrow_down:
smdebug/core/logger.py	70.83% <0.00%> (-5.56%)	:arrow_down:
smdebug/core/access_layer/s3.py	91.54% <0.00%> (-4.23%)	:arrow_down:
smdebug/core/reader.py	85.18% <0.00%> (-3.71%)	:arrow_down:
... and 25 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 314c091...42117b0. Read the comment docs.