Added DeepSeek R1 + DeepSeek V3 benchmark

[serialx]

I'd like to share the result of DeepSeek R1 architect + DeepSeek V3 editor benchmark results: It's 59.1%. Near the performance of o1 but at the fractional cost of $6.33! Half of R1+Sonnet.

- dirname: 2025-01-25-13-53-23--deepseek-r1-v3
  test_cases: 225
  model: deepseek/deepseek-reasoner
  edit_format: architect
  commit_hash: b276d48-dirty
  editor_model: deepseek/deepseek-chat
  editor_edit_format: editor-diff
  pass_rate_1: 30.7
  pass_rate_2: 59.1
  pass_num_1: 69
  pass_num_2: 133
  percent_cases_well_formed: 100.0
  error_outputs: 13
  num_malformed_responses: 0
  num_with_malformed_responses: 0
  user_asks: 388
  lazy_comments: 1
  syntax_errors: 0
  indentation_errors: 0
  exhausted_context_windows: 0
  test_timeouts: 4
  total_tests: 225
  command: aider --model deepseek/deepseek-reasoner
  date: 2025-01-25
  versions: 0.72.3.dev
  seconds_per_case: 949.4
  total_cost: 6.3330

Note: The dirty commit simply occurred because I've added DEEPSEEK_API_KEY env variable to the docker launch parameter. Otherwise it's a genuine clean copy of the code.

[CLAassistant]

All committers have signed the CLA.