Some updates to the svs viewer, splitting the image viewing and agent modules.

Recently I wanted to get the SVS Viewer base demo online, but I ran into a few issues. If I keep developing the agent module on the SVS Viewer structural framework, the base gets increasingly complex, and eventually I won't be able to adjust the original image viewing module anymore.

After researching some materials and consulting GPT, I decided to switch the base's viewer module to pathTogather's collaborative image viewing base, and make the agent analysis/navigation module a separate plugin. This way, on one hand, the pathTogather base can quickly produce a demo and be released, and on the other hand, updates won't easily affect the navi module.

The idea is roughly to simulate the agent as a shared user, using the pathTogather base just like a regular user would for viewing images, annotating, and switching perspectives. This also lets me open a new SDK module on pathTogather, so that in the future all programs can simulate a collaborator's perspective to mark up or capture screenshots.

Plus, since I've always intended for pathTogather to handle some of the region selection work for spatial transcriptomics, this framework is fairly solid. From here, it's just a matter of waiting until it's deployed and seeing how it performs.