Keep duplicate "<file> 2" copies out of commits

Something on this machine generates byte-identical copies named "<file>
2.go", "<file> 3.py" and so on, across build output, virtualenvs and
source trees alike — 1,277 of them were cleaned out of ~/Documents/GitHub
in one pass, 190 MB worth, including ten Python modules shadowing their
originals in another repo. Two landed in internal/agent during a session
here and broke the build with redeclaration errors, which is how they
were noticed at all.

The cause is still unknown. It is not iCloud: ~/Documents is not synced,
the symlink in CloudDocs points the other way, and no file carries an
iCloud attribute.

This only keeps them out of commits. It does not stop the build breaking,
because the file is still on disk and the compiler still reads it — so
the comment says where to look when a build fails for no reason.

The pattern is broad enough to catch a legitimately named "Chapter 2.md",
which is a fair trade in a Go repository that has no such file and cannot
tolerate the ones it does get.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

Author: torrmal

.gitignore

@@ -30,3 +30,13 @@ go.work.sum
 # Editor/IDE
 # .idea/
 # .vscode/
+
+# Duplicate copies named "<file> 2.go", "<file> 3.py" and so on. Something
+# on this machine generates them across build output, virtualenvs and
+# source trees alike — 1,277 identical ones were cleaned out of
+# ~/Documents/GitHub in one pass. A stale copy beside a real .go file
+# breaks the build with redeclaration errors, so this keeps them out of
+# commits at least. It does not stop the build breaking: the file is still
+# on disk and the compiler still reads it. `find . -name '* [0-9]*'` finds
+# them when a build fails for no apparent reason.
+* [0-9]*