feat: benchmark util for assessing performance

[Nischay-Pro]

Description

This PR adds a new benchmarking script to quickly check the performance of flexmeasures and allows the devs to check the performance of the code.

To use, simply have flexmeasures configured and activated in your local environment, and run

python3 benchmark.py

located in the ci folder of the root directory.

---

• I agree to contribute to the project under Apache 2 License.
• To the best of my knowledge, the proposed patch is not based on code under GPL or other license that is incompatible with FlexMeasures

[nhoening]

I tested it and it works fine. Thanks!

Could you round the measured duration to a shorter representation (e.g. 2 after-comma digits)?

Maybe @Flix6x has ideas for the commands:

• Under which circumstances do these work? (I believe this assumes our production db dump)?
• Could we maintain the list of sensors outside of the main code, or above in a handy-to-edit list?
• What other commands would be useful, e.g. add schedule for-storage

[Nischay-Pro]

I tested it and it works fine. Thanks!

Could you round the measured duration to a shorter representation (e.g. 2 after-comma digits)?

Maybe @Flix6x has ideas for the commands:

* Under which circumstances do these work? (I believe this assumes our production db dump)?

* Could we maintain the list of sensors outside of the main code, or above in a handy-to-edit list?

* What other commands would be useful, e.g. `add schedule for-storage`

• Currently production db-dump
• That actually looks good. Something similar to toy tutorial, with it's own dataset
• Yes. That would be helpful.

[Flix6x]

I would like it if the sensor ID (for me, one is enough) could be passed when calling the benchmark script, so the script can be used on any FlexMeasures database. For me, it would then make sense to also allow passing a custom start, duration and max_iterations (possibly with the current values as defaults). I do like the benchmark showing results for several durations, but a single sensor is all I need.

[nhoening]

@Flix6x I don't agree this PR was ready. Let's re-open it.

It should not call the CLI as that takes too long by itself. I have not found the best way to benchmark otherwise yet. Maybe API or CLI or calling into functions directly. But as you in one of my commits, the client is not a good way either IMO.

[Flix6x]

Apologies. I reverted the merge commit on the main branch: 8ed96e1.

[nhoening]

How can we re-open a PR?

[nhoening]

Oh I see #949 is the new one for this.