[yfwmaniish]

Fixes winstonjs/winston#2365.

What

Both code paths in `Splatter` separate "meta" (extra arguments beyond what the format string's tokens consume) from "splat" by calling `.splice()` directly on `info[SPLAT]`:

```js
// _splat() - tokens path
const metas = extraSplat < 0
? splat.splice(extraSplat, -1 * extraSplat)
: [];
...
info.message = util.format(msg, ...splat); // splat has now been mutated

// transform() - no-tokens path
const metas = splat.length > 1
? splat.splice(0)
: splat;
```

`info[SPLAT]` is an array reference. winston shallow-clones `info` once per transport (each transport's own format pipeline gets its own `info` object), but a shallow clone still shares the same `info[SPLAT]` array across all of them. So the first transport's `.splice()` mutates that shared array in place - by the time the second transport's format pipeline runs `splat()` over its own `info` object, `info[SPLAT]` is already truncated (or emptied entirely), and the meta winston attached to the log call is silently gone.

Reproduces exactly as described in the issue: two `Console` transports with `format.combine(format.splat(), format.json())`, only the first one gets `extraData`.

Fix

Both call sites now read the same values via `.slice()` instead of `.splice()` - same values extracted, but the original array (and thus `info[SPLAT]`) is left untouched. _splat()'s local splat (used for util.format) is likewise derived via .slice() instead of relying on the post-splice, already-mutated array.

Testing

Added two regression tests (tokens path and no-tokens path), each simulating two transports as two separate `info` objects sharing one `SPLAT` array reference - reusing the same `info` object across both transform() calls wouldn't actually catch this bug (a stale property surviving on the same object would mask it), so the test mirrors winston's real shallow-clone-per-transport shape.

• Negative control: reverted just `splat.js`, kept the tests - both fail with the meta reading `undefined` on the second transport, matching the real bug. Restored.
• Full suite: `mocha test/*.test.js` - 171/171 passing.
• `eslint` clean on both changed files (verified against the git-staged/LF-normalized content; this Windows checkout has `core.autocrlf` converting the whole working tree to CRLF locally, which is unrelated noise - confirmed by checking an untouched file too).

Separate from #362 (which fixes a different bug - primitive metas getting char-spread via `Object.assign`, tracked as #109). This PR doesn't touch that code path.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.