Add t2v multiple line queues and v2v multi-file queues

[bfasenfest]

Refactored the code a bit so the logic about computing latents from video frames was separated out from the other logic.

Then added a basic queue using a deque.

For text to video, if you add prompts on new lines it will process them in a batch, similar to how the "Prompts from file or textbox" works in the normal txt2img tab.

For example, entering the following into the prompt input, will set up a queue where they process in sequence:
dancing man
dancing frog
dancing octopus

This means you can set up a big list of prompts to run over a long period of time instead of having to start them individually. In this mode, changing the batch count will create a new batch for each video (so batch count 2 for the above would make 6 total videos, 2 of each).

For vid2vid, the video input now accepts multiple files. If you drop in multiple files it will process them with the given prompt in sequence. Again, quite helpful if you are testing out different outputs with the same prompt and want to upscale all at the same time.

[nuclearsugar]

This fork has been really useful, especially having the Vid2Vid queue. Unfortunately it doesn't seem to function with the latest A1111 update. Any chance of a fix?

[nuclearsugar]

Reverted back to A1111 v1.4.1 and it functions smoothly again.