Specifiy code upload location for processing jobs

[marcelgwerder]

Describe the feature you'd like
For estimators that are based on the Framework class and upload local code, it is possible to provide a code_location argument that specifies the location of the uploaded code. This helps to avoid cluttering the root "directory" of the s3 bucket because everything can go under a prefix/directory of choice. The same should be possible for processing jobs.

How would this feature be used? Please describe.
The developer provides a code_location argument when instantiating the ScriptProcessor. The value of that argument is then put in between bucket and job name when creating the upload path. By default the value can be empty, meaning there is no breaking change necessary and this can be released anytime.

Describe alternatives you've considered
It would be best to align this with how it is done for training jobs, so the above solution should work out fine.

[liujiaorr]

Can anyone give a update if the request has been processed or still in need?

[meilipfi]

We had similar issues when using processing jobs and would benefit from a solution where we can specify where the code will be uploaded in the S3 bucket