The results.json was not generated because vegeta was not found.
For some reason the error was caught and silenced which made it harder
to debug.

1) adding `ipykernel` fixed:
```
jupyter_client.kernelspec.NoSuchKernel: No such kernel named python3
```

2) adding `ipython_genutils` fixed:
```
ModuleNotFoundError: No module named 'ipython_genutils'
```

3) bumping the `nbconvert` version fixed:
```
ImportError: cannot import name 'contextfilter' from 'jinja2'
(/home/flood/.local/lib/python3.11/site-packages/jinja2/__init__.py)
```

This was tested within a Docker container, built with:
```
docker build --tag paradigmxyz/flood:latest .
```
Ran with:
```
docker run --rm --volume ${PWD}/output:/output paradigmxyz/flood:latest \
eth_getBlockByNumber NODE:8545 --output /output/
```

Add existing report example link in report generation section

Fix parquet file download error because of a missing curl binary.
This was happening on the `eth_getBalance` test for instance.
Note that flood swallowed the exception making the error harder to spot.

Update the CI to also test against eth_getBalance which triggers a
parquet file download.
Also removes the redundant node2, this was a workaround as flood used to
crash with only one node.

Also unpin the patch version of Python and address a small DRY on the
Vegeta version.

…into eth_call_perf

Add possibility to manipulate start/end block