fix(ai-config): intern model in gptel-switch-backend (modeline hang)

cj/gptel-switch-backend set gptel-model to the raw completing-read string. gptel's modeline code calls symbolp on gptel-model and signals wrong-type-argument on a string, which surfaces as a redisplay hang (reachable from C-; a B). The sibling command cj/gptel-change-model already interns. This one didn't.

I added a pure cj/gptel--model-to-symbol helper (mirroring cj/gptel--model-to-string) and route the model through it before the setq. The existing switch-backend test asserted the buggy string value. It now asserts a symbol plus an explicit symbolp guard.

1 files changed, 4 insertions, 1 deletions

diff --git a/tests/test-ai-config-gptel-commands.el b/tests/test-ai-config-gptel-commands.el
index b87c4975..371a75cc 100644
--- a/tests/test-ai-config-gptel-commands.el
+++ b/tests/test-ai-config-gptel-commands.el
@@ -77,7 +77,10 @@
                (lambda (fmt &rest args) (setq msg (apply #'format fmt args)))))
       (cj/gptel-switch-backend))
     (should (eq gptel-backend 'anthropic-backend))
-    (should (equal gptel-model "claude-opus"))
+    ;; gptel-model must be a symbol, not the raw completing-read string:
+    ;; gptel's modeline calls `symbolp' on it and hangs redisplay otherwise.
+    (should (symbolp gptel-model))
+    (should (eq gptel-model 'claude-opus))
     (should (string-match-p "Anthropic - Claude" msg))))
 
 (ert-deftest test-ai-config-switch-backend-error-invalid-choice ()