Skip corrupt pages instead of aborting the whole conversion

[NilsLeo]

A single unreadable or zero-byte page raised RuntimeError and failed the entire book, even when every other page was fine. This skips the bad page and carries on, while keeping a genuine environment error (a missing Pillow decoder) fatal so real setup problems still surface. It also guards against a missing cover, so a damaged archive with no usable cover page no longer crashes with a NoneType error.

How to reproduce / test

A comic with one zero-byte page and one unreadable/garbage page mixed in with valid pages triggers both fatal branches this PR fixes:

Download: https://drive.google.com/file/d/1rz38SkzJLcZ3D_laJCdC8XYo5nRg9Ghi/view?usp=drive_link

Save it next to where you run the commands (referred to below as 1399.cbz). The archive is flat (images at the root) with page-003-EMPTY.jpg (0 bytes) and page-005-GARBAGE.jpg (non-image bytes) between valid pages.

The easiest way to A/B test any KCC branch without a local setup is kcc-run, which clones a given owner/repo:ref into a throwaway container at runtime:

export DOCKER_DEFAULT_PLATFORM=linux/amd64   # Apple Silicon; no-op on Intel/AMD
kcc() { docker run --platform linux/amd64 --rm -v "$PWD:/work" -w /work ghcr.io/nilsleo/kcc-run "$@"; }

1. Reproduce the crash on upstream master (one corrupt page fails the whole book):

kcc ciromattia/kcc:master -p KV -f EPUB -o out 1399.cbz

→ aborts with RuntimeError: Image file .../page-005-GARBAGE.jpg is corrupted. at detectSuboptimalProcessing; no EPUB produced.

2. Same file on this branch — skips the bad pages and converts the rest:

kcc NilsLeo/kcc:up/image-resilience -p KV -f EPUB -o out 1399.cbz

→ logs

WARNING: Skipping corrupt image .../page-005-GARBAGE.jpg (cannot identify image file ...)
WARNING: Skipping corrupt (empty) image .../page-003-EMPTY.jpg

then writes a valid out/1399.epub from the remaining pages.

(Or clone this branch and run python kcc-c2e.py -p KV -f EPUB -o out 1399.cbz directly.)

[NilsLeo]

@axu2 this is constructed, but ive encountered files like this several times that have these sort of corrupted pages