fix(params): mark test executable and catch float(result) conversion errors

Two follow-up fixes on top of the ConfigurationStepEvalError refactor:

- Set the executable bit on tests/test_data_model_safe_evaluator.py so
  ruff stops failing with EXE001 (shebang present but file not
  executable). The file was created with a shebang matching other
  tests; only the git mode bit was missing.

- Extend the outer except in compute_parameters to also catch
  ValueError and TypeError. The Par(float(result), ...) call on the
  non-eval path can raise either when result is a string that came
  from a values/Bitmask lookup but is not numerically convertible, or
  when the dict lookup returned an unexpected type. Without these,
  conversion errors escaped both try blocks and could abort the
  configuration-step load. SyntaxError, NameError, KeyError, and
  StopIteration remain for defensive coverage of the surrounding
  code paths.

Signed-off-by: Yash Goel <yashvardhan664@gmail.com>

Author: yashhzd

ardupilot_methodic_configurator/backend_filesystem_configuration_steps.py

@@ -249,7 +249,7 @@ def log_parameter_error(
                     destination[filename] = ParDict()
                 change_reason = _(parameter_info["Change Reason"]) if parameter_info["Change Reason"] else ""
                 destination[filename][parameter] = Par(float(result), change_reason)
-            except (SyntaxError, NameError, KeyError, StopIteration) as _e:
+            except (SyntaxError, NameError, KeyError, StopIteration, ValueError, TypeError) as _e:
                 error_msg = _(
                     "In file '{self.configuration_steps_filename}': '{filename}' {parameter_type} "
                     "parameter '{parameter}' could not be computed: {_e}"