Improve Auto3DSeg for MLFLow tracking, waiting, and cleanup

[drbeh]

This PR

• adds mlflow tracking uri to the payload and user parameter section,
• specify a timeout for job waiting,
• remove inference dataset from training, and
• fix the mismatched job_id variable in cleanup (change to infer_job_id)

[mingxin-zheng]

Hi @drbeh , all these notebooks follow a certain format so that the integration test developed by @yiheng-wang-nv will run smoothly. Rearranging functions can potentially break the tests. I'll pass it to @yiheng-wang-nv for the review. Thanks!

[drbeh]

Note: cleanup issue was already addressed in other merged PR, so it is removed after resolving conflicts.

[drbeh]

Hi @drbeh , all these notebooks follow a certain format so that the integration test developed by @yiheng-wang-nv will run smoothly. Rearranging functions can potentially break the tests. I'll pass it to @yiheng-wang-nv for the review. Thanks!

Thanks @mingxin-zheng! Hopefullychanging the place of a helper function doesn't break the integration tests but @yiheng-wang-nv is there any reason not have the helper function at the top?

[Cadeww000]

✌️😎

[yiheng-wang-nv]

Hi @drbeh , all these notebooks follow a certain format so that the integration test developed by @yiheng-wang-nv will run smoothly. Rearranging functions can potentially break the tests. I'll pass it to @yiheng-wang-nv for the review. Thanks!

Thanks @mingxin-zheng! Hopefullychanging the place of a helper function doesn't break the integration tests but @yiheng-wang-nv is there any reason not have the helper function at the top?

Hi @drbeh , that's the suggestion from @zephyrie . Related comment: https://github.com/NVIDIA/monai-cloud-api/pull/34#discussion_r1454362283

[drbeh]

Thanks @yiheng-wang-nv! I changed back the location of the helper function, so I can merge this.