fix: Tensor not detached in debug mode (ecmwf#279)

## Description
Output tensor was not detached when printing it or writing it to a trace
file, side effect of ecmwf#263.


***As a contributor to the Anemoi framework, please ensure that your
changes include unit tests, updates to any affected dependencies and
documentation, and have been tested in a parallel setting (i.e., with
multiple GPUs). As a reviewer, you are also responsible for verifying
these aspects and requesting changes if they are not adequately
addressed. For guidelines about those please refer to
https://anemoi.readthedocs.io/en/latest/***

By opening this pull request, I affirm that all authors agree to the
[Contributor License
Agreement.](https://github.com/ecmwf/codex/blob/main/Legal/contributor_license_agreement.md)

Author: gmertes

src/anemoi/inference/runner.py

@@ -642,11 +642,15 @@ def forecast(
                     new_state["fields"][self.checkpoint.output_tensor_index_to_variable[i]] = output[:, i]
 
             if (s == 0 and self.verbosity > 0) or self.verbosity > 1:
-                self._print_output_tensor("Output tensor", output)
+                self._print_output_tensor("Output tensor", output.cpu().numpy())
 
             if self.trace:
                 self.trace.write_output_tensor(
-                    date, s, output, self.checkpoint.output_tensor_index_to_variable, self.checkpoint.timestep
+                    date,
+                    s,
+                    output.cpu().numpy(),
+                    self.checkpoint.output_tensor_index_to_variable,
+                    self.checkpoint.timestep,
                 )
 
             yield new_state