chore: add April 2026 data

[akalyuzhnyi]

Hello @patrickhulce

I'm updating the data for April 2026. Could you do a review, please?

[patrickhulce]

thanks so much @akalyuzhnyi ! wow everything got so much worse though, like 6x4x.

@rviscomi or @paulirish any chance you know of a major change to HTTPArchive hardware or test run in April? Or is this a bug on our end we should investigate further?

[rviscomi]

📟 @pmeenan @tunetheweb

[pmeenan]

Nothing changed with the HTTP Archive that I'm aware of on the collection side. The instance types, configuration and images haven't been touched in close to a year and the only agent updates have been around updating feature names and wappalyzer detections.

[tunetheweb]

wow everything got so much worse though, like 6x.

Can you explain this more? Not sure what I’m looking at to see that level of change.

[patrickhulce]

Thanks very much all! I'll look more deeply into this on our end then. Initial findings...

Metric	March 2026	April 2026	Δ	%
totalExecutionTime (sum)	71.65 B ms	288.01 B ms	+216.36 B ms	+302%
totalScripts (sum)	172.65 M	681.40 M	+508.74 M	+295%
totalOccurrences (sum)	71.67 M	69.30 M	−2.36 M	−3.3%
Mean exec time per script	415 ms	423 ms	+8 ms	+1.85%
Mean exec time per occurrence	1,000 ms	4,156 ms	+3,156 ms	+316%

So pretty much all scripts duplicated 4x in this PR, but looking at specific Lighthouse reports from the April sample on HTTPArchive nothing obvious jumps out (no duplicated URLs, no URLs that aren't part of the page, etc). Either something in the report structure changed or we've gone awry in our querying in this one.

[patrickhulce]

@akalyuzhnyi I think something went wrong with the querying process here. Do your intermediate third_party_web tables have duplicate script entries?

I can't verify these results against the actual HTTPArchive data in BigQuery.