[dash,youtube] Download live from start to end (#888)

* Add option `--live-from-start` to enable downloading live videos from start
* Add key `is_from_start` in formats to identify formats (of live videos) that downloads from start
* [dash] Create protocol `http_dash_segments_generator` that allows a function to be passed instead of fragments
* [fragment] Allow multiple live dash formats to download simultaneously
* [youtube] Implement fragment re-fetching for the live dash formats
* [youtube] Re-extract dash manifest every 5 hours (manifest expires in 6hrs)
* [postprocessor/ffmpeg] Add `FFmpegFixupDuplicateMoovPP` to fixup duplicated moov atoms

Known issue: Ctrl+C doesn't work on Windows when downloading multiple formats

Closes #1521
Authored by: nao20010128nao, pukkandan

README.md

@@ -88,6 +88,7 @@ yt-dlp is a [youtube-dl](https://github.com/ytdl-org/youtube-dl) fork based on t
     * Redirect channel's home URL automatically to `/video` to preserve the old behaviour
     * `255kbps` audio is extracted (if available) from youtube music when premium cookies are given
     * Youtube music Albums, channels etc can be downloaded ([except self-uploaded music](https://github.com/yt-dlp/yt-dlp/issues/723))
+    * Download livestreams from the start using `--live-from-start`
 
 * **Cookies from browser**: Cookies can be automatically extracted from all major web browsers using `--cookies-from-browser BROWSER[:PROFILE]`
 
@@ -340,6 +341,10 @@ You can also fork the project on github and run your fork's [build workflow](.gi
     --flat-playlist                  Do not extract the videos of a playlist,
                                      only list them
     --no-flat-playlist               Extract the videos of a playlist
+    --live-from-start                Download livestreams from the start.
+                                     Currently only supported for YouTube
+    --no-live-from-start             Download livestreams from the current
+                                     time (default)
     --wait-for-video MIN[-MAX]       Wait for scheduled streams to become
                                      available. Pass the minimum number of
                                      seconds (or range) to wait between retries
@@ -1585,7 +1590,7 @@ The following extractors use this feature:
 * `skip`: `hls` or `dash` (or both) to skip download of the respective manifests
 * `player_client`: Clients to extract video data from. The main clients are `web`, `android`, `ios`, `mweb`. These also have `_music`, `_embedded`, `_agegate`, and `_creator` variants (Eg: `web_embedded`) (`mweb` has only `_agegate`). By default, `android,web` is used, but the agegate and creator variants are added as required for age-gated videos. Similarly the music variants are added for `music.youtube.com` urls. You can also use `all` to use all the clients, and `default` for the default clients.
 * `player_skip`: Skip some network requests that are generally needed for robust extraction. One or more of `configs` (skip client configs), `webpage` (skip initial webpage), `js` (skip js player). While these options can help reduce the number of requests needed or avoid some rate-limiting, they could cause some issues. See [#860](https://github.com/yt-dlp/yt-dlp/pull/860) for more details
-* `include_live_dash`: Include live dash formats (These formats don't download properly)
+* `include_live_dash`: Include live dash formats even without `--live-from-start` (These formats don't download properly)
 * `comment_sort`: `top` or `new` (default) - choose comment sorting mode (on YouTube's side)
 * `max_comments`: Limit the amount of comments to gather. Comma-separated list of integers representing `max-comments,max-parents,max-replies,max-replies-per-thread`. Default is `all,all,all,all`.
     * E.g. `all,all,1000,10` will get a maximum of 1000 replies total, with up to 10 replies per thread. `1000,all,100` will get a maximum of 1000 comments, with a maximum of 100 replies total.