Remove licensedcode data from SCIO docker image

[AyanSinhaMahapatra]

Scancode.io docker image contains both the scancode-toolkit licenses index and all the license data files, which increases the image size (and number of resources in that image) by a lot.
We probably only need the index and not the data files, as reindex might not be required in the majority of cases.
See aboutcode-org/scancode-toolkit#3459 where we are planning to split the scancode wheels, and this is related.

[tdruez]

@AyanSinhaMahapatra Do we need to run a scancode command in the Dockerfile before removing the data files to ensure that the index is properly built?

Also, could you clarify which directory should be removed?

[AyanSinhaMahapatra]

Do we need to run a scancode command in the Dockerfile before removing the data files to ensure that the index is properly built?

Yeah we need to run scancode-reindex-licenses to ensure that the index is properly built, from a environment with scancode installed. See how we do it for building linux app archives to release: https://github.com/aboutcode-org/scancode-toolkit/blob/develop/etc/release/scancode-create-release-app-linux.sh#L67

Also we can safely remove opt/scancodeio/scanpipe/tests/ from the docker image too without issues.

As mentioned in the referenced issues, we will soon package the license index and the data separately in two wheels to reduce the size of wheels and app archives, and then the license-data would not be present in a installed scancode-toolkit wheel by default (but the index will be).