#4057: fix: sanitize tool call IDs for Azure OpenAI

Azure OpenAI enforces a 40-character limit on tool_calls[].id fields, which causes HTTP 400 errors when tool call IDs exceed this limit. Add Azure OpenAI detection (via provider name containing "azure") and enable tool call ID sanitization for Azure providers. <!-- greptile_comment --> <h2>Greptile Overview</h2> <h3>Greptile Summary</h3> This PR updates `resolveTranscriptPolicy` (`src/agents/transcript-policy.ts`) to detect “Azure OpenAI” providers and enable transcript/tool-call sanitization to avoid Azure’s 40-character `tool_calls[].id` limit. The policy object returned here is consumed by the session history/message sanitizers (e.g. `sanitizeSessionMessagesImages`), so toggles like `sanitizeToolCallIds` and `toolCallIdMode` directly affect whether outbound requests have tool-call IDs rewritten into a stricter format. <h3>Confidence Score: 2/5</h3> - Not safe to merge as-is because the main intended behavior change appears not to take effect for Azure OpenAI. - The PR adds an `isAzureOpenAi` flag and includes it in `sanitizeToolCallIds`/`needsNonImageSanitize`, but the returned policy still gates `sanitizeToolCallIds` behind `!isOpenAi`, so Azure (which uses OpenAI APIs) likely won’t have tool-call IDs sanitized and will continue to hit 400s. Azure detection logic is also heuristic and may miss real Azure configs depending on provider naming/normalization. - src/agents/transcript-policy.ts <!-- greptile_other_comments_section --> _{(2/5) Greptile learns from your feedback when you react with thumbs up/down!} **Context used:** - Context from `dashboard` - CLAUDE.md ([source](https://app.greptile.com/review/custom-context?memory=fd949e91-5c3a-4ab5-90a1-cbe184fd6ce8)) - Context from `dashboard` - AGENTS.md ([source](https://app.greptile.com/review/custom-context?memory=0d0c8278-ef8e-4d6c-ab21-f5527e322f13)) <!-- /greptile_comment -->