fix(settings): restore missing large PR handling config for pr_description (#2234)

* fix(settings): restore missing large PR handling config for pr_description

  The `enable_large_pr_handling`, `max_ai_calls`, and `async_ai_calls`
  keys were removed from the `[pr_description]` section in
  `settings/configuration.toml`, but `pr_description.py` still accesses
  `get_settings().pr_description.enable_large_pr_handling` at runtime.

  When the key is absent, an exception is raised inside
  `_prepare_prediction()` before `get_pr_diff` is even called.
  `retry_with_fallback_models()` catches it silently and logs
  "Failed to generate prediction", causing all configured models to
  exhaust in milliseconds with no actual API call ever made.

  Fixes #2230

* fix(settings): restore inline_file_summary config for pr_description

`inline_file_summary` was missing from the `[pr_description]` section
in `settings/configuration.toml`. `pr_description.py:130` accesses
`get_settings().pr_description.inline_file_summary` at runtime, and
the missing key causes an exception that silently fails the describe
command.

Fixes #2230

* fix(settings): remove Pro-only label from large PR handling config

---------

Co-authored-by: Hank <pupu721@gmail.com>

Author: Hank076

pr_agent/settings/configuration.toml

@@ -107,10 +107,14 @@ publish_description_as_comment_persistent=true
 enable_semantic_files_types=true
 collapsible_file_list='adaptive' # true, false, 'adaptive'
 collapsible_file_list_threshold=6
+inline_file_summary=false # false, true, 'table'
 # markers
 use_description_markers=false
 include_generated_by_header=true
 #custom_labels = ['Bug fix', 'Tests', 'Bug fix with tests', 'Enhancement', 'Documentation', 'Other']
+enable_large_pr_handling=true
+max_ai_calls=4
+async_ai_calls=true
 
 [pr_questions] # /ask #
 enable_help_text=false