Auth issues: Multi-channel account (delegated session id) and multi-account cookies (SESSION_INDEX) are not available as we don't have ytcfg

Some options (only use if authenticated):

• https://www.youtube.com/getDatasyncIdsEndpoint:

  • will get correct sync id for account
  • will give data sync ids for all accounts
  • does not give session index (don't think it is possible to derive since the list changes)

• v1/account/accounts_list:

  • will give data sync ids of all channels
  • can derive session id from "accountSigninToken" url
  • does not tell us what the account the cookies are for.
  • does not appear to give other accounts that are logged in?

• https://www.youtube.com/getAccountSwitcherEndpoint:

  • Parts of response are similar to above
  • Shows all accounts & channels

Curious as to how the non-web based clients get this data. OAuth?

For interest: ytmusicapi requires the user to pass the datasyncid (and session index): https://ytmusicapi.readthedocs.io/en/latest/usage.html#brand-accounts

Random note: datasyncid/accountStateToken/DELEGATED_SESSION_ID/onBehalfOfUser/pageIdToken are pretty much all the same thing in this context...

Edit: I've gone with the third option. It seems like these "hacky" endpoints as what YouTube calls them are as they are called.

Another issue:
There appears to be only one working STS value (i.e., using an older player version with it's STS value will not work). Unless we have a way to get the STS or player version in an api-only way, we are stuck with only using non-js based player clients.

Some changes:

• When the player url fails to be extracted by other means, we fallback to extracting it from the iframe_api page.
• We do not extract the player_url/player js/sts if the client does not require it
  • This is configured by setting REQUIRES_JS_PLAYER in the hard-coded ytcfgs. Default is True.
  • We can override this decision globally by using player_js=no_require/require arg (TODO: better name)
• if a client requires sts/player js and we fail to extract it, that client will be skipped.
  • TODO: maybe don't want this behaviour since in non-signature related cases it doesn't matter if we have sts/player js
• player url can be also extracted from player_ytcfg (e.g. embedded, web_music)
• we can disable the initial webpage download with player_skip=webpage.
  • TODO: do we want this under player_skip=configs?

Currently, to enable a pure api-only mode for YoutubeIE, we can pass --extractor-args youtube:player_skip=webpage,configs;player_js=no_require.

Slightly unrelated: you can get 429nd for downloading too many subtitles in a short amount of time. But, the 429 appears to only affect subtitle downloads.

Some extra things to fix:
MP -> OLAK playlist resolve is broken
fix/improve tests
generalise extract_webpage and extract_response in some way (might make for another PR though)

Superseded by #1122