V108 QUICK FINITE-FAULT LOCAL DEPTH FIX

Base
- Built from V107.

Problem fixed
- Quick finite-fault placement previously assigned one fixed depth to each patch center.
- When the patch width was increased, the up-dip/down-dip patch edges did not get shallower/deeper.
- This made the visible model behave as if the effective dip became smaller.

What changed
- Each source cell inside a quick finite-fault patch now gets its own depth.
- Patch center depth still follows: origin_depth + rel_d * fault_width * sin(dip).
- Inside each patch, local down-dip offset adds: projected_offset_km * tan(dip).
- The injection height is recomputed per source cell using slip, dip, and that local depth.

Result
- Increasing patch width now increases the depth difference between patch edges.
- The requested dip is preserved better instead of effectively flattening.
