studio: address review feedback

install_python_stack.py:
- Print uv error output on failure for debuggability
- Refactor pip_install() to use early return after uv success,
  removing duplicated pip command path

setup.sh:
- Guard nvidia-smi command substitution with || true so it does
  not abort the script under set -euo pipefail when nvidia-smi
  fails (e.g., containerized environments, driver quirks)
- Read all GPU compute capabilities and deduplicate, so
  mixed-GPU hosts get kernels built for all present architectures
  instead of only the first GPU

studio/install_python_stack.py

@@ -186,13 +186,14 @@ def pip_install(
                 stdout = subprocess.PIPE,
                 stderr = subprocess.STDOUT,
             )
-            if result.returncode != 0:
-                print(_red(f"   uv failed, falling back to pip..."))
-                pip_cmd = _build_pip_cmd(args) + constraint_args + req_args
-                run(label, pip_cmd)
-        else:
-            pip_cmd = _build_pip_cmd(args) + constraint_args + req_args
-            run(label, pip_cmd)
+            if result.returncode == 0:
+                return
+            print(_red(f"   uv failed, falling back to pip..."))
+            if result.stdout:
+                print(result.stdout.decode(errors = "replace"))
+
+        pip_cmd = _build_pip_cmd(args) + constraint_args + req_args
+        run(label, pip_cmd)
     finally:
         if actual_req is not None and actual_req != req:
             actual_req.unlink(missing_ok = True)