Add model settings for DeepSeek R1 and V3 family on Azure AI Foundry

[Cyberfilo]

Summary

Adds four entries to `aider/resources/model-settings.yml` for the DeepSeek family hosted on Azure AI Foundry (`azure_ai/` provider — distinct from the native `deepseek/` provider):

Model	weak_model_name
`azure_ai/deepseek-r1` (reasoner)	`azure_ai/deepseek-v3`
`azure_ai/deepseek-v3` (chat)	(no weak, defaults)
`azure_ai/deepseek-v3-0324` (chat, pinned)	(no weak, defaults)
`azure_ai/deepseek-v3.2` (chat, newer)	(no weak, defaults)

These models exist in litellm's pricing JSON (aider's upstream source for model metadata via `ModelInfoManager.MODEL_INFO_URL`) but were missing project-side settings. Without them, aider falls back to defaults that don't match the DeepSeek family conventions (`reminder: sys`, `caches_by_default: true`, `use_temperature: false` for the reasoner).

Settings shape

Mirrors the existing native `deepseek/deepseek-reasoner` and `deepseek/deepseek-chat` blocks exactly:

• `edit_format: diff` (the DeepSeek family's preferred edit grammar)
• `use_repo_map: true`
• `examples_as_sys_msg: true`
• `caches_by_default: true` (DeepSeek-style implicit caching)
• `reminder: sys` for chat variants
• `use_temperature: false` + `max_tokens: 64000` for the reasoner; `max_tokens: 8192` for chat variants
• Reasoner routes to v3 for weak/editor model duties

Test plan

• YAML validity: `python -c "import yaml; yaml.safe_load(open('aider/resources/model-settings.yml'))"` parses cleanly.
• Diff is purely additive: 40 lines in model-settings.yml + 1 line in HISTORY.md.