[release/4.0] Improve unique directory generation for temp files

[github-actions]

Backport of #7520 to release/4.0

/cc @ericstj

Customer Impact

For each model opened ML.NET would leave behind a temp folder which would slowly accumulate in the temp directory. It also had an algorithm for temp folder creation which would linearly degrade with each new folder.

Testing

Tested fix that the folder is deleted and the creation algorithm does not degrade linearly if folders happen to be left behind because it uses GetTempFileName to produce significantly random folder name candidates.

Risk

This fix doesn't clean up past folders, we should release-note that users may want to do this manually.

[ericstj]

Need to fix the build, will first do that here -- darc-release/4.0-59b01119-72fa-4ec9-8595-f64b50c7a93e - #7470

[tarekgh]

LGTM. The build issue is not directly related to the change. Eric is fixing that.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 68.89%. Comparing base (7b6eccf) to head (e0f447e).
⚠️ Report is 2 commits behind head on release/4.0.

Additional details and impacted files

@@             Coverage Diff              @@
##           release/4.0    #7528   +/-   ##
============================================
  Coverage        68.89%   68.89%           
============================================
  Files             1470     1470           
  Lines           274081   274082    +1     
  Branches         28405    28405           
============================================
+ Hits            188836   188840    +4     
  Misses           77928    77928           
+ Partials          7317     7314    -3     

Flag	Coverage Δ
Debug	68.89% <100.00%> (+6.25%)	⬆️
production	63.30% <100.00%> (+0.66%)	⬆️
test	89.42% <ø> (∅)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/Microsoft.ML.Core/Data/Repository.cs	80.00% <100.00%> (+0.06%)	⬆️

... and 4 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[ericstj]

/azp run MachineLearning-CI

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).