I assume the timing patterns and amounts of data would likely be distinct between SSH and web. "Normal" SSH usage would mostly consist of much lighter packets, such as user keystrokes and terminal screenfuls of text. Typing tiny commands and getting a few kilobytes of output. SSH file transfers happen occasionally, sometimes with a large bulk of data.
Active web browsing requires downloading a crapton of files with wildly different sizes and sporadic timings between them.
Add normal user interaction, API requests, ad cycles, long video streams that won't max out all bandwidth, all happening at once across multiple tabs. The client also sends much more data with each TLS handshake and all those HTTP headers.
This could probably be masked by deliberately filling idle periods with garbage data just to appear as a stable data stream both ways.
Forget using a real web browser over an SSH proxy. What using elinks on a remote host with ssh? I bet somebody using elinks across ssh is virtually indistinguishable from somebody using a text editor.
Active web browsing requires downloading a crapton of files with wildly different sizes and sporadic timings between them. Add normal user interaction, API requests, ad cycles, long video streams that won't max out all bandwidth, all happening at once across multiple tabs. The client also sends much more data with each TLS handshake and all those HTTP headers.
This could probably be masked by deliberately filling idle periods with garbage data just to appear as a stable data stream both ways.