Video Input resume_from_timestring continues till video max_frames but repeatedly generates same image from previous batch

Using latest version of Automatic1111 with deforum-for-automatic1111-webui.

Running with video input on Google Colab, session got cut off, finished 3433 frames out of 4185.

Checked "resume_from_timestring", and entered the correct timestring, and the same settings, the run finished when I noticed the newly generated frames, although in order and up to 4185, all show the same generated image which is based on the last successful image from the first session.

These are the final frames of the first session, next to the first frames of the second session, next to the video frames:

As you can see, the frames of the second session are a copy of the last "known" frame 3433, so the video frames were not processed.

Settings file:
[20230101090619_settings.txt](https://github.com/deforum-art/deforum-for-automatic1111-webui/files/10346314/20230101090619_settings.txt)

I would've tried to debug it myself but still lacking knowledge, I would gladly help if someone can point me in the right direction :)

Start by reproducing the report with the linked 20230101090619_settings.txt settings on a resumed video-input run, comparing frames after frame 3433 with the source video. Trace the resume path and verify that resumed frames process the corresponding video inputs rather than repeating the last generated image; done means frames through 4185 match the continued sequence.

Written by the indexing model from the issue text.