Stable release of CM v1.5.3 and Collective Knowledge v3 supporting MLPerf inference benchmark v3.1+

Stable release of the MLCommons CM automation language v1.5.3 and Collective Knowledge v3 supporting MLPerf inference benchmark v3.1+ and our community submissions.

CM v1.5.2: minor improvements for MLPerf inference and Artifact Evaluation

• fixed minor bug with auto-initializion of cmind for cmind.error and cmind.halt for external scripts
• added --min to cm find repo xyz --min to print path to CM xyz repo without any extra info
  Needed for artifact evaluation at ACM MICRO'23
• added cm where repo xyz to print path to xyz CM repo without any extra info
  Needed for artifact evaluation at ACM MICRO'23

CM v1.5.1: bug fixes

• fixed a bug when merging dictionaries in variations (detected when running MLPerf inference with GPT-J): #858

CM v1.5.0: bug fixes

Fixed a serious (though rare) bug in indexing when mixing entries with UIDs and aliases.

V1.4.1: added "cme" binary shortcut for "cm run experiment"

a few updates (#790)

- added cme binary as a shortcut for "cm run experiment"
- improved TinyMLPerf tutorials

V1.4.0 new "cmr" binary and support for TinyMLPerf

• added cmr binary as a shortcut for "cm run script"
• fixed minor bug with repo/automation/artifact detection in the current path with "."
• various improvements in CM-MLOps repository to support TinyMLPerf

V1.3.0: indexing and silent mode turned on by default

• Turned on artifact indexing by default
  (can be turned off by setting CM_INDEX to "no", "off" or "false")
• Turned on --silent mode in "cm run script" by default
  Can be turned off via --verbose or -v flags
• Fixed duplicate version detection for Python packages
• Added --new_tags for "cm copy" to add new tags to newly created artifacts
• Added --new_tags for "cm add" to add new tags to newly created artifacts
• Added basic check for forbidden artifact names such as "."

V1.2.2: bug fixes

Fixed minor bug during cm detect repo (turn off indexing).

V1.2.1

V1.2.0: Major update: transparent indexing of all artifacts to speed up search and scripts by ~10..50x.

It is currently OFF by default for further testing.
You can turn it on and test by setting this environment variable: "CM_INDEX=yes|on|true".

V1.2.1: Fixed Bug in indexing to avoid returning duplicate artifacts when wildcards are used.

MLCommons CM v1.1.6

Minor fixes and updates: https://github.com/mlcommons/ck/blob/master/cm/CHANGES.md#v116