BTreeMap: fix pointer provenance rules in underfullness

[ssomers]

Continuing on #78480, and for readability, and possibly for performance: avoid aliasing when handling underfull nodes, and consolidate the code doing that. In particular:

• Avoid the rather explicit aliasing for internal nodes in remove_kv_tracking.
• Climb down to the root to handle underfull nodes using a reborrowed handle, rather than one copied with ptr::read, before resuming on the leaf level.
• Integrate the code tracking leaf edge position into the functions performing changes, rather than bolting it on.

r? @Mark-Simulacrum

[Mark-Simulacrum]

r=me on the last commit (I did not review the first two, but I believe they have been previously reviewed, so that shouldn't be a concern). Ping me when those are rebased away after the previous PRs land, please.

[bors]

☔ The latest upstream changes (presumably #78889) made this pull request unmergeable. Please resolve the merge conflicts.

Note that reviewers usually do not review pull requests until merge conflicts are resolved! Once you resolve the conflicts, you should change the labels applied by bors to indicate that your PR is ready for review. Post this as a comment to change the labels:

@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

[ssomers]

I tweaked the name and comments of choose_balancing_context → choose_parent_kv

[ssomers]

@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

Also, github doesn't seem to process commits.

[Mark-Simulacrum]

First commit indeed seems to be on master, not sure why GitHub is showing it. Thanks!

@bors r+

[bors]

📌 Commit b6c26c4 has been approved by Mark-Simulacrum

[ssomers]

The commit you approved is the one GitHub showed me too, but I had (tried to) push a rebase on the newest master. Now GitHub sees that rebase commit, but somehow it's still targeting the old master. While the branch of this PR looks perfectly normal in GitHub itself: one commit on a recent master.

I doubt bors will attempt to take off in this state, so trying to wake up GitHub by pushing an even fresher rebase.

[ssomers]

Aha, just one commit instead of 35 now.

[Mark-Simulacrum]

@bors r+

[bors]

📌 Commit 5a129ac35cd7bdb64f762db4eeade48223ebf331 has been approved by Mark-Simulacrum

[bors]

☔ The latest upstream changes (presumably #78956) made this pull request unmergeable. Please resolve the merge conflicts.

Note that reviewers usually do not review pull requests until merge conflicts are resolved! Once you resolve the conflicts, you should change the labels applied by bors to indicate that your PR is ready for review. Post this as a comment to change the labels:

@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

[ssomers]

Oh right, fix_right_edge moved to its own file.
@rustbot modify labels: +S-waiting-on-review -S-waiting-on-author

[Mark-Simulacrum]

@bors r+ rollup=never

[bors]

📌 Commit 4cfa5bd has been approved by Mark-Simulacrum

[bors]

⌛ Testing commit 4cfa5bd with merge 3e5a681469d8559e2bb5f9f384487c960db92d72...

[bors]

💔 Test failed - checks-actions

[Mark-Simulacrum]

@bors retry

[bors]

⌛ Testing commit 4cfa5bd with merge 1cf9258ce60a03dbbb4589d2df424dc1e11db52c...

[bors]

💔 Test failed - checks-actions

[Mark-Simulacrum]

@bors retry

[bors]

⌛ Testing commit 4cfa5bd with merge f5230fb...

[bors]

☀️ Test successful - checks-actions
Approved by: Mark-Simulacrum
Pushing f5230fb to master...