[icons] Consider as written in TypeScript

[oliviertassinari]

A continuation on #15984 (comment). It seems that migrating the icons from .js to a .tsx extension wouldn't make sense considering that we generate all these files as well as their .d.ts definitions. The current approach is more efficient. However, I believe we can consider these files are written in TypeScript.

Before

$ github-linguist
94.80%  JavaScript
5.19%   TypeScript
0.01%   HTML
0.00%   Shell

After

$ github-linguist
54.48%  TypeScript
45.50%  JavaScript
0.01%   HTML
0.00%   Shell

I have considered using linguist-generated until I saw:

be ignored for the repository's language statistics and hidden by default in diffs

https://docs.github.com/en/github/administering-a-repository/customizing-how-changed-files-appear-on-github

[mui-pr-bot]

No bundle size changes comparing 993a623...3318b7a

Generated by 🚫 dangerJS against 3318b7a

[eps1lon]

I mean it's not TS so it's not TS.

[oliviertassinari]

I mean it's not TS so it's not TS.

@eps1lon I can't disagree, JS is not TS 😆. Here is the underlying problem I'm going after:

• Short term: the reported JavaScript vs TypeScript ratio doesn't seem to draw an accurate representation of the usage of TypeScript in the codebase: 5% of TypeScript sources?!
• Long term: we are progressively moving toward a codebase fully written in TypeScript, once we will have fully migrated the components, we would still have GitHub reports a majority of JavaScript files which will be still misleading for new users. Why? because the icons account for half the files in the repository.

The only alternative I can think of is to generate .tsx rather than .js files in the icons script. However, this seems completely artificial, as we would skip tslint on these files to save time (no need to lint the, it's always the same pattern).

[oliviertassinari]

This change assumes that people look at the language states and that it has an influence in their decisions for picking a dependency, that when they see 5%, they will search for "TypeScript" in the issue history to better understand the state of it. They will land on #15984 and blindly upvote the issue without spending a lot of time looking closely at the coverage.

[oliviertassinari]

has an influence in their decisions for picking a dependency

Motivation number 4 in or survey results: https://material-ui.com/blog/2020-developer-survey-results/#6-what-are-your-key-criteria-when-choosing-a-ui-library

[oliviertassinari]

Question 6 in the survey is about user acquisition, which is not to be confused with current users that experience a lot less TypeScript issues in the survey of 2020 than 2019 :)

[eps1lon]

But the stats are wrong either way. At least previously we didn't intentionally manipulate them.

[oliviertassinari]

Moving to #22188