Refactor LAMBKIN

[hidmic]

Proposed changes

Closes #17, #24, #34. Partially addresses #18 and #19. To call this a patch is an understatement. This is almost a different project. The overall architecture and spirit of LAMBKIN remains, but the implementation has changed substantially:

• lambkin, the Python package, has been generalized and modularized to support further extension. A form of descriptive DSL within RobotFramework allows benchmark configuration, but many hooks are available for user override. Benchmark data access is now behind a Python API, enabling programmatic use (one step away from Jupyter Notebooks as a report #26).
• Benchmark ROS packages have been migrated to match the new API. Overall structure remains the same.
• Provisioning, and containerization in particular, has been reworked to be easier to develop and release packages, both in this repository and downstream, and to make better use of existing tooling instead of reinventing the wheel.

One notable omission are CI workflows. I intend to add them in an immediate follow-up PR.

Type of change

• 🐛 Bugfix (change which fixes an issue)
• 🚀 Feature (change which adds functionality)
• 📚 Documentation (change which fixes or extends documentation)

💥 Breaking change! This a complete rework of this project. No way (and no need) to merge this incrementally while maintaining backwards compatibility.

Checklist

• Lint and unit tests (if any) pass locally with my changes
• I have added tests that prove my fix is effective or that my feature works
• I have added necessary documentation (if appropriate)

Additional comments

A good place to start looking into this is documentation (still scarce, I have plenty to write).

Each commit is a conceptually separate set of changes, which is not to say each commit is functional on its own. They are not. I abandoned that precept a long time ago. That said, I'm open to split this PR in whatever way makes it easier to review.

[hidmic]

One important question not yet answered above is why, why this change, why this huge PR. I must say I should've known better and I should've organized work so as to avoid a 10k LOC patch. To be honest, when I started with #35 I was not sure where I was heading. As the patch evolved, so did my understanding of the problem and the limitations of the existing implementation. Today, with this architecture and the mental model I have of it, LAMBKIN can sustain growth (of features, use cases, collaborators) in a way our humble proof of concept couldn't.

[hidmic]

FYI @agalbachicar @glpuga.

[nahueespinosa]

@hidmic This is more than amazing!

I'm going to review this by parts. Until now:

• I followed the "getting started" guidelines to run local benchmarks (SLAM Toolbox Magazino and parameterized TUM).
• I did a first pass on the general directory structure.
• I did a first pass on the user-facing .robot scripts.

[francocipollone]

Great job indeed,

I just did one pass. I will proceed to run different benchmarks with different options and see if I identify a bug or misbehavior.

[nahueespinosa]

@hidmic LGTM! This is amazing.

I haven't gone through every single line of code, but I don't see any point in delaying this PR any longer.
Future documentation and feature work will reveal improvement points (if any), but this is as solid as it gets in my opinion.

[glpuga]