For terminal-centric workflows, one ends up needing to open dozens of shells throughout the day.
Starting one command, realizing you need info from a previous command, often the fastest way to get that info is to open another terminal and start typing and the copy-pasting the relevant piece of cmd. Or needing the output from another command that's scrolled up in a window. In general you end up with a mosaic of tiles each of which is holding key information. So you inevitably open a new tab to make use of the information in the other 3 or 4 windows.
Shell startup is mainly not about waiting for a few terminals in abstraction, but about the way each shell is just a small piece of a longer live workflow and waiting a second for a shell so that you can fill in a piece of a command feels terrible. I think nushell will help cut down on this unpiped inefficiency, but it's usually much simpler to manually glue bash commands together than to parse text output and pipe.
Starting one command, realizing you need info from a previous command, often the fastest way to get that info is to open another terminal and start typing and the copy-pasting the relevant piece of cmd. Or needing the output from another command that's scrolled up in a window. In general you end up with a mosaic of tiles each of which is holding key information. So you inevitably open a new tab to make use of the information in the other 3 or 4 windows.
Shell startup is mainly not about waiting for a few terminals in abstraction, but about the way each shell is just a small piece of a longer live workflow and waiting a second for a shell so that you can fill in a piece of a command feels terrible. I think nushell will help cut down on this unpiped inefficiency, but it's usually much simpler to manually glue bash commands together than to parse text output and pipe.