LRV proxies should be refused by name, pointing at the VID sibling

Owner hit this testing #129: passing an LRV_ file to the harness (or app) fails with the generic dimension-mismatch refusal. By #107's refuse-by-name standard, the app should recognize the LRV_ prefix / proxy signature and say plainly that it is the camera's low resolution proxy and the full recording is the VID_ file beside it. Options for the ruling: (a) name it and point at the sibling in the #117 alert; (b) with #123/#127's capture-set resolution, auto-open the VID sibling instead (zero-config, but opens a different file than asked). Also: the harness could fail fast on LRV_ inputs with a one-line hint instead of two 45 s timeouts.

Start by locating the harness/app file-input flow that now emits the generic dimension-mismatch refusal for LRV_ files. Follow the existing refuse-by-name handling referenced in the issue thread (#107/#117) and add an LRV_ case that clearly explains it is a low-res proxy and points to the VID_ sibling. Check capture-set behavior described in #123/#127 to decide whether to keep warning-only or auto-open the VID_ file. Verify by running the harness/app with an LRV_ input and confirming a fast single-line hint instead of two 45s timeouts.

由索引模型根據 Issue 內容生成。