Feature: Per-segment audio effects DSP preset selector for dub pipeline

[4shil]

Problem

The dub pipeline currently has no way to customize audio treatment per segment. All segments use the same mastering chain (apply_mastering with fixed highpass + compressor + reverb). Different scenes benefit from different audio treatment: podcast dialogue needs heavy compression, action scenes need presence boosts, cinematic scenes need reverb.

Proposed Solution

Add a per-segment effect_preset field to DubSegment that lets users choose from 6 broadcast-grade DSP presets (broadcast, cinematic, podcast, warm, bright, raw). This leverages existing, well-tested code in audio_dsp.py (apply_effects_chain, EFFECT_PRESETS, list_effect_presets()) that was previously only used in standalone TTS generation.

Scope

• Backend API + Frontend state management only (no UI component in this PR)
• Non-breaking: defaults to broadcast preset, existing behavior unchanged
• No database schema changes
• No model changes

Files Changed

• backend/api/routers/engines.py — new GET /engines/effects/presets route
• backend/api/schemas.py — EffectPresetEntry + EffectPresetsResponse
• backend/schemas/requests.py — effect_preset field + Pydantic validator on DubSegment
• backend/api/routers/dub_generate.py — apply effect chain in TTS hot loop
• backend/services/batched_tts.py — apply effect chain in batch TTS; effect_preset in SegmentSpec
• backend/api/routers/generation.py — effect_preset form field in standalone /generate
• backend/services/incremental.py — effect_preset in fingerprint fields
• frontend/src/api/engines.ts — EffectPreset type + fetchEffectPresets()
• frontend/src/api/types.ts — DubSegment type with effect_preset
• frontend/src/store/dubSlice.ts — segmentEffectPresets state + actions
• tests/test_effects_chain.py — 12 unit tests (pure functions, no GPU)

Verification

uv run pytest tests/test_effects_chain.py -v