For a 0.8.6 release

[ablaom]

193

Release notes:

• Add logger option to TunedModel wrapper, for logging internal model evaluations to an ML tracking platform. Default should be nothing for no logging (#193). The logger must support asynchronous messaging if TunedModel has acceleration != CPU1(). Such support is planned for MLflow integration via MLJFlow.jl; see https://github.com/JuliaAI/MLJFlow.jl/pull/41 #193

cc @pebeto

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 90.00000% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 87.55%. Comparing base (1dd32b1) to head (2294da4). Report is 33 commits behind head on master.

Files	Patch %	Lines
src/tuned_models.jl	90.00%	1 Missing :warning:

Additional details and impacted files

```diff @@ Coverage Diff @@ ## master #218 +/- ## ========================================== + Coverage 87.53% 87.55% +0.01% ========================================== Files 13 13 Lines 666 667 +1 ========================================== + Hits 583 584 +1 Misses 83 83 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.