Script runners small enhancements

[brian-mulier-p]

Issue description

See comments

[brian-mulier-p]

PEBBLE:
{{workingDir}}
{{outputDir}}
{{bucketPath}}

ENV:
WORKING_DIR
OUTPUT_DIR
BUCKET_PATH

Rename (gcs/s3/azure)Bucket to bucket

Move uploadInputFiles to CommandsWrapper (with additionalVars precomputed before scriptRunner.run())

Open questions:

• outputDir -> useless ? Since on cloud providers it is set to the same value as workingDir
• in DOCKER & PROCESS runners everything put in {{outputDir}} are put in the outputFiles. However such operation is costly on cloud providers so we don't want to do this behaviour

To add in doc' -> limitation due to glob patterns hard to handle in remote script runners (we can't list easily files from remote buckets). Currently, remote runners only support hardcoded relative paths to file. We could extend it to prefix but not much further

[loicmathieu]

For cloud runner, outputDir should be created inside the bucket and all files in this path (via a single prefix query) should be downloaded as outputFiles.

[loicmathieu]

These enhancements has been done, I'll close this one.