Add cache at Step level

[plaguss]

Description

This PR implements cache at step level.

Previously, we computed a signature for a pipeline, and when this signature changed, we recomputed everything.
Now the idea is to compute the signature per step, and once the signature has changed, only recompute the steps whose signature (or preceding) has changed. So for a pipeline A -> B -> C, if B step changes, we will recompute only B and C, but we will start with the data we had from A.

New cases we control with this change:

• Some failure in a Pipeline during the computation (say we push Ctrl+c), we can restart from where we left.
• If we have a pipeline a >> b >> c >> d and we change a step (say c), we will only recompute c and d.
• Control the cache use per step level. We have an argument use_cache at the _Step level, when set to False, the cache won't be used from that step onwards, even if the pipeline remains the same.

step_b = MyStep(
    name="step_b",
    input_batch_size=10,
    use_cache=False,
)

Note: This has an impact in how we read the previous serialized parquet files, if any step's use_cache is set to False, for a pipeline that hasn't changed, we won't read the previous serialized content.

Closes #651

[codspeed-hq]

CodSpeed Performance Report

Merging #766 will degrade performances by 25.09%

_{Comparing cache-per-step (3ea2b85) with develop (e027f99)}

Summary

❌ 1 (👁 1) regressions

Benchmarks breakdown

	Benchmark	develop	cache-per-step	Change
👁	test_cache_time	398.9 ms	532.6 ms	-25.09%

[github-actions]

Documentation for this PR has been built. You can view it at: https://distilabel.argilla.io/pr-766/