[UnknownSuperficialNight]

I'm not sure if this is something we want to even commit while the rewrite is happening in #901 but ill add the PR in case feel free to close if it's irrelevant in the future or if we need updating ping me here

Problem

Player::get_pos() is wrong whenever the speed is not 1.0.

Speed does not change the samples, it only reports a scaled sample rate.
TrackPosition sat after Speed in the chain and divided its sample count
by that scaled rate, so it measured elapsed playback time instead of the
position in the media. At 0.5x the position advanced 1s per second while
the audio advanced 0.5s per second, and changing the speed during
playback made it wrong in other ways too.

Player::try_seek was also affected: Speed::try_seek multiplies the
target by the speed factor, so try_seek(30s) at 0.5x landed at 15s of
the media.

Changes

• Player::append: track_position() now comes before speed(), so
  samples are counted at the source's real sample rate. It still sits
  inside pausable(), so silence played while paused is not counted.
• SeekOrder::attempt: the target is divided by the current speed to
  cancel the multiplication in Speed::try_seek. The division uses f64 and
  falls back to the unscaled position if the result is out of range.
• Updated the .inner() chains in the periodic_access closure for the
  new adapter order and documented the order.
• Docs: rewrote TrackPosition::get_pos, updated the speed() and
  set_speed docs and the get_pos/try_seek docs, replaced an outdated
  example in SpatialPlayer::get_pos, and linked the related items.
• Added a test for the track_position_before_speed ordering.

Behaviour

• get_pos() is the position in the media, independent of speed.
• try_seek(pos) seeks to pos in the media at any speed.
• At speed 1.0 nothing changes.

Testing

• Played a track at 0.5x and 1.5x: get_pos() advances at half and 1.5x
  the rate of real time, and stays continuous when the speed changes
  during playback.
• try_seek(30s) lands at 30s at 0.5x, 1.0x and 2.0x.
• cargo test and cargo doc

Crude Diagram

This is a crude diagram showing what changed I was going to use the mermaid diagram renderer on GitHub, but its sucks so here is a PNG instead.

This does not account for pause and all that in the diagram It's simply just the core idea basically.

Additional notes

1. This branch is forked from my old AGC branch as the main branch at the moment still has Clicking/glitching sound when set_speed is higher or lower than 1.0 using opus #917 and or something else after that branch added the issue, so I just based it off my own for working AGC branch.
2. I also made this fix simply because I needed it fixed for my program as I'm releasing soon just figured I would submit a PR in case.

closes #837