blocks-import-script: add --markdown flag and remove --csv-output flag

[siddarthkay]

Summary

This PR adds --markdown-output support to blocks import script.
This PR also removes --csv-output flag.

This option can be used like this

python scripts/block-import-stats.py
short-benchmark/20241208T172603_9207eaf/blocks-import-benchmark.csv
short-benchmark/20241222T012247_aba9b582/blocks-import-benchmark.csv 
--markdown-output README.md

The output is a nice markdown table :

blocks-import-benchmark.csv vs blocks-import-benchmark.csv

Block Range	BPS Baseline	BPS Contender	TPS Baseline	TPS Contender	Time Baseline	Time Contender	BPS Diff	TPS Diff	Time Diff
(20000001, 20111112]	25.62	24.98	4176.33	4075.37	1h9m29s	1h11m34s	-2.63%	-2.63%	2.84%
(20111112, 20222223]	25.91	26.95	4237.79	4408.99	1h13m53s	1h11m3s	4.01%	4.01%	-3.84%
(20222223, 20333334]	26.60	27.69	4314.13	4490.81	1h7m0s	1h4m26s	4.05%	4.05%	-3.87%
(20333334, 20444445]	29.13	30.55	4586.04	4811.19	1h5m47s	1h2m47s	4.83%	4.83%	-4.59%
(20444445, 20555556]	29.12	29.79	4483.73	4588.42	1h0m59s	59m38s	2.29%	2.29%	-2.22%
(20555556, 20666667]	29.62	30.58	4472.00	4619.38	1h4m33s	1h2m32s	3.26%	3.26%	-3.14%
(20666667, 20777778]	30.50	31.44	4797.46	4947.33	58m15s	56m32s	3.08%	3.08%	-2.97%
(20777778, 20888889]	30.99	32.87	4969.59	5273.27	1h1m50s	58m18s	6.08%	6.08%	-5.72%
(20888889, 21000001]	30.81	32.98	5080.18	5438.98	1h2m16s	58m16s	7.00%	7.00%	-6.53%

Summary

Metric	Value
Total Blocks	991808
Baseline Time	9h44m5s
Contender Time	9h25m10s
Time Difference	-18m55s
Time Difference %	-3.24%

Legend

• BPS Diff: Blocks per second difference (+)
• TPS Diff: Transactions per second difference
• Time Diff: Time to process difference (-)

(+) = more is better, (-) = less is better

[yakimant]

(20000001, 20111112] - different brackets?

[siddarthkay]

(20000001, 20111112] - different brackets?

In python : The brackets ( and ] in the output (20000001, 20111112] represent an "interval notation".

( means exclusive (does not include) the first number
] means inclusive (includes) the last number

[arnetheduck]

my preference would broadly be that we cut down on the output formats to as few as possible - ie either csv or markdown - there's no point maintaining a bunch of formats that don't get used.

[siddarthkay]

I like csv as an output format.
However judging by what you want as part of Readme on Github, the markdown table output seems like what we should keep as an output format.

[arnetheduck]

whatever is minimally legible .. I'd even be fine with dumping the regular text output into triple-quotes or a table or a csv or any variation thereof. It doesn't greatly matter as long as it doesn't require multiple clicks to get to. The fewer the clicks and scrolling the better.

In terms of importance, I think the "main" readme that lists all benchmarks in a big table is the most important one - this is where we can get a quick overview whether a particular commit slowed something down. Having to drill down into each folder takes ages.

[jakubgs]

You can keep CSV and then just use a converter tool - one of many, like csv2md - and just convert it afterwards.

[jakubgs]

The markdown formatting code could use refactoring but it's good enough for now.