rendering: depth of field + motion blur (post FX pack 2)

[shadow-kernel]

Context

Both are listed in the verified rendering gaps for horror (cinematic focus cues, impact/turn smearing) and were never consciously deferred — without a ticket they'd silently fall out of the catalog. The v2.7 post-processing framework makes them straightforward follow-ups, and motion vectors already exist engine-wide (DLSS FG pass).

Goal

Add depth of field (near/far blur with focus distance/range) and per-pixel motion blur (reusing the existing motion-vector target) as passes in the post-processing chain, both scriptable for cinematic moments.

Acceptance Criteria

• DoF pass: focus distance + range + max CoC, gather-based blur, no visible haloing on the template scene
• Motion blur pass sampling the existing RG16F motion-vector target, intensity clamped and configurable
• Both toggleable/parameterized via the post-FX API from game scripts (e.g. focus pull on a document)
• Sane interaction with DLSS SR/FG documented (apply post-upscale; FG mvec tagging unaffected)
• Combined GPU cost budget: < 1.5 ms at 1440p on the reference GPU (numbers in PR)

Technical Notes

Run after tonemap-independent passes in the chain established by the framework issue; DoF needs depth (already bound for DLSS). Start with half-res CoC gather + upsample.

Dependencies

• rendering: post-processing framework (pass chain between scene and UI)