visualize inter-job back-pressure rate in dashboard

[BugenZhao]

Since #13830 and #14446, we're now able to acquire the back-pressure rate in the Streaming Fragment Graph of the dashboard.

However, it's pretty common that there are multiple streaming jobs in the cluster, where the slowness of some one impacts others. This requires us to have a broader view of the back-pressure among different jobs to identify the root cause, which is not possible yet.

One possible implementation might be integrating the data into the Dependency Graph page. It would also be beneficial to reimagine the Streaming Fragment Graph page, enabling users to dynamically zoom in and out to view all streaming jobs in the cluster from both macro and micro perspectives.

[BugenZhao]

I think this is done by @kwannoel in #18280?