Fetching external benchmarks data

Currently, all data is assumed to be cloned from the benchmarks-data repository or downloaded from a GH release. However, we need to provide additional data that is (a) under a problematic license or (b) too large to fetch from GitHub. In particular, this problem appears in SeBS-Flow.

We should have a fetch system that:

• Runs after initializing benchmarks-data
• Runs all custom "fetch" scripts in the benchmarks-data.
• Clearly communicates this to the user - how much we fetch and from where (we should put size expectations to not surprise users).
• Adds a hidden marker to skip refetching.

Good place to install this feature: https://github.com/spcl/serverless-benchmarks/blob/master/sebs/utils.py#L737

CC: @lolamaspoli

Start in sebs/utils.py around line 737 and inspect how benchmarks-data is initialized, then review the custom fetch scripts in the benchmarks-data repository. Define the fetch step, user-facing size and source information, and the hidden marker behavior. Done means initialization runs the scripts once and skips them on subsequent runs when the marker is present.