fix(pixel-inference): reject negative or out-of-range temperature in prepare

[vaibhavsrv]

Why this matters

In ods/extensions/services/pixel-inference/app/main.py, _prepare(payload, grant) checks inference request fields against the allowed FIELDS set and enforces message formatting, but does not validate the type or numerical bounds of temperature. When a client passes a negative temperature, a value exceeding the model runtime cap (> 2.0), or a non-numeric/boolean value (such as true or a string), the unvalidated value passes through to downstream inference runtimes, triggering unhandled backend exceptions, process abortion, or HTTP 500 crashes.

This change adds a surgical validation block to _prepare before computing the token limit. It verifies that temperature is an int or float (excluding bool) within the canonical [0.0, 2.0] range, raising ShareError(400, 'invalid_temperature') otherwise. Existing valid temperatures, default behavior, and model parameter fallbacks remain completely unchanged.

Validation

• Tested baseline reproduction against unpatched main.py: passing {"temperature": -1.0}, {"temperature": 2.5}, or {"temperature": True} bypassed validation and was forwarded without rejection.
• Tested post-fix behavior: negative, out-of-range, and boolean/string temperatures are rejected with HTTP 400 and code invalid_temperature; valid values in [0.0, 2.0] succeed identically.
• Telemetry statement: "Pixel-inference suites: 4 passed. New-test Ruff, py_compile, and diff checks pass; new regression wired into Linux CI."

Overlap check

• fix(healthcheck): support bracketed IPv6 targets and strip brackets before TCP connect #5875 (merged): Fixed bracketed IPv6 parsing in healthcheck.py.
• fix(pixel-settings): reject active token limits exceeding provider capacity #5696 (merged): Clamped active token limits in pixel_settings/contract.py.
• fix(pixel-edge): accumulate trailing buffered content and synthesize fallback finish #5608 (merged): Handled trailing SSE buffer accumulation in pixel_edge.py.
• perf(pixel): scan stream terminator incrementally instead of recopying 2MB tail #5647 (merged): Optimized streaming terminator scanning in pixel_gateway.
  This PR operates exclusively on incoming parameter validation in pixel-inference/app/main.py and does not overlap with or duplicate any of these changes.

Risk / AI disclosure

AI-assisted investigation, implementation, and unit regression. This strengthens an inference parameter contract check and does not modify running infrastructure, model files, or container configurations. Independent human review and platform qualification remain gates.

Follow-up integration evidence

Composed cleanly on top of #5569, #5871, and #5875 at HEAD without conflicts. Production and test diffs passed together; adjacent pixel-agent and pixel-settings checks remain intact.
Backlog composition was local-only (production/test diffs, excluding workflow/Makefile wiring); it is not an upstream merge or independent human approval. Declared live-review gates remain open.