[codex] Add token budget context feature (#27438)

## Why

The model should be able to see bounded context-window budget metadata
when the `token_budget` feature is enabled. The full-window message is
only injected with full context, while normal turns get a smaller
follow-up only when reported usage first crosses a budget threshold.

## What changed

- Added the `TokenBudget` feature flag.
- Added `<token_budget>` developer fragments for full context-window
metadata and current-window remaining tokens.
- Inserted the threshold message during normal turn handling by
comparing token usage before and after sampling, avoiding persistent
threshold bookkeeping.
- Added core integration coverage for full-context-only metadata and
25/50/75 percent threshold messages.

## Verification

- `just test -p codex-core token_budget`
- `git diff --check`

codex-rs/core/src/compact_remote.rs

@@ -249,6 +249,7 @@ async fn run_remote_compact_task_inner_impl(
             turn_metadata_header.as_deref(),
         )
         .await?;
+    let new_window_id = sess.advance_auto_compact_window_id().await;
     new_history = process_compacted_history(
         sess.as_ref(),
         turn_context.as_ref(),
@@ -264,7 +265,7 @@ async fn run_remote_compact_task_inner_impl(
     let compacted_item = CompactedItem {
         message: String::new(),
         replacement_history: Some(new_history.clone()),
-        window_id: None,
+        window_id: Some(new_window_id),
     };
     // Install is the semantic boundary where the compact endpoint's output becomes live
     // thread history. Keep it distinct from the later inference request so the reducer can