Create single_channel.py

[PrachiUkey]

feat: switch to single‑channel render optimization; update save utils for grayscale outputs

refactor: use first render channel for loss, safe save for 1‑channel images, new obj name

fix: depth‑channel training → single‑channel synthetic optimization; handle grayscale saving

[PrachiUkey]

depth‑channel training → single‑channel synthetic optimization; handle grayscale saving

[PrachiUkey]

Replaced RGB + alpha image optimization with single-channel synthetic data optimization. The renderer now outputs only the first channel (previously interpreted as “depth”), and the loss is computed on this channel for all views. Added safe image saving for single‑channel outputs and updated result OBJ saving.