Impact of third-party render blocking scripts

[mukeshpanchal27]

Fixes #30

Query results

Row	percentile	client	host	num_requests
1	10	desktop	first party	8
2	10	desktop	third party	4
3	25	desktop	first party	16
4	25	desktop	third party	10
5	50	desktop	first party	32
6	50	desktop	third party	22
7	75	desktop	first party	56
8	75	desktop	third party	42
9	90	desktop	first party	86
10	90	desktop	third party	68
11	100	desktop	first party	3320
12	100	desktop	third party	11578
13	10	mobile	first party	6
14	10	mobile	third party	3
15	25	mobile	first party	12
16	25	mobile	third party	6
17	50	mobile	first party	24
18	50	mobile	third party	16
19	75	mobile	first party	45
20	75	mobile	third party	34
21	90	mobile	first party	74
22	90	mobile	third party	56
23	100	mobile	first party	2174
24	100	mobile	third party	2324

Based on June 2022 dataset.

For the query use below references:

• https://almanac.httparchive.org/en/2022/javascript#requests
• https://github.com/HTTPArchive/almanac.httparchive.org/blob/main/sql/2022/javascript/requests_by_3p.sql
• https://github.com/HTTPArchive/almanac.httparchive.org/blob/main/sql/2022/javascript/render_blocking_javascript.sql

[felixarntz]

@mukeshpanchal27 Left a few comments below. Overall this looks good, but I'd like for someone from the HTTP Archive team with more experience in those queries to have a look too.

Can you please provide a reference in the PR description for what the query is based on (https://almanac.httparchive.org/en/2022/javascript#requests, or more specifically https://github.com/HTTPArchive/almanac.httparchive.org/blob/main/sql/2022/javascript/requests_by_3p.sql)?

Regardless of whether the query is entirely correct like this or not, I just ran it (consumes 3 TB, so not too bad) and posted the results on the PR description.

[felixarntz]

Oh I see now, here we use pages too, and this part wouldn't be in the technologies table. So maybe you're right and we need to keep using that table. What is the check here for?

[mukeshpanchal27]

As per Percent of pages using render-blocking JavaScript, We can parse the pages.payload to get render blocking script counts.

[kt-12]

The main goal here is to study render-blocking third-party scripts.
However, the sub-query i.e (httparchive.almanac.third_parties), brings up 1p and 3p scripts, both render-blocking and non-render blocking.

We are making this study on a site where at least one script is render-blocking, but we can't say through this query if the third-party script is part of the render-blocking script.

cc: @mukeshpanchal27 @felixarntz

[kt-12]

@mukeshpanchal27 I am convinced this query will return the distribution of all the third parties and first parties (both render blocking and non-render blocking) for a site with at least one render-blocking script.

This won't return the distribution of render-blocking third-party and render-blocking first-party scripts.
Please find my explanation and example in this comment

I am finding it hard to explain in this PR.
cc: @felixarntz