ScriptV2: NPM library codebase

[macobo]

Depends on #5449. If that gets rejected, I will rebase accordingly.

This PR is the start of work for creating a new NPM library for our tracker script off of the same tracker codebase.

import { init, track } from '@plausible/tracker' # or some other library name

init({ 
  domain: "plausible.io"
})

It:

• Has typescript types (via a type declaration file)
• Shares majority of tests with plausible-web (needs a rebase after Update playwright #5432 is merged)
• Has protections against double-init/track-before-init, throwing in that case

Work left for the future:

• Release process via CI
• Renaming the library, MIT license, CHANGELOG.md and README.md
• Commonjs variant. Work on this was in this PR but was hard to test for.

The PR also:

• Refactors tests to be less susceptible to timing issues (specifically around pageview trigger timing)
• Removes features from compile.js that allowed limiting what to compile. Not needed with the new compile being faster.

[github-actions]

Analyzed 1026 tracker script variants for size changes.
The following tables summarize the results, with comparison with the baseline version in parentheses.

Main variants:

	Brotli	Gzip	Uncompressed
npm_package/plausible.js (new variant)	2328B	2671B	6275B
plausible-web.js	2270B (-11B / -0.5%)	2594B (-11B / -0.4%)	6103B (-20B / -0.3%)

Important legacy variants:

	Brotli	Gzip	Uncompressed
plausible.js	1140B (+0B / 0%)	1328B (-5B / -0.4%)	2883B (+0B / 0%)
plausible.hash.js	1103B (+0B / 0%)	1294B (-4B / -0.3%)	2769B (+0B / 0%)
plausible.pageview-props.tagged-events.js	1842B (+0B / 0%)	2144B (-5B / -0.2%)	4797B (+0B / 0%)
plausible.file-downloads.hash.pageview-props.revenue.js	1666B (+9B / +0.5%)	1954B (-4B / -0.2%)	4143B (+4B / +0.1%)
plausible.compat.exclusions.file-downloads.outbound-links.pageview-props.revenue.tagged-events.js	2250B (+2B / +0.1%)	2655B (-5B / -0.2%)	5941B (+4B / +0.1%)

Summary:

	Brotli	Gzip	Uncompressed
Largest variant (plausible.compat.exclusions.file-downloads.outbound-links.pageview-props.revenue.tagged-events.js)	2250B (+2B / +0.1%)	2655B (-5B / -0.2%)	5941B (+4B / +0.1%)
Max change (plausible.compat.local.manual.pageview-props.tagged-events.js)	1602B (+14B / +0.9%)	1900B (-3B / -0.2%)	4120B (+4B / +0.1%)
Min change (plausible.compat.exclusions.local.manual.outbound-links.pageview-props.js)	1419B (-13B / -0.9%)	1712B (-8B / -0.5%)	3495B (+4B / +0.1%)
Median change	1696B (+0B / 0%)	2006.5B (-4B / -0.2%)	4322B (+4B / +0.1%)

In total, 378 variants brotli size increased and 256 variants brotli size decreased.

[apata]

I like this! I appreciate the shared tests and that the npm module is tested at all. Lots of removed lines and small nuisances fixed.

[apata]

Less repetition in tests, very nice!

[Daniel15]

If you search Google for "Plausible npm", the old package is still returned, and the new one isn't in the results at all. Most people are still downloading the old script rather than the new one, based on npm download stats. Do you have plans to mark the old one as deprecated at some point?