Use blocking reads for setup_terminal

[lhecker]

As per James Holderness' suggestion, we actually don't
need ESC timeouts here. In this phase of the app, escape
chars are very unlikely to be from keyboard input.

[DHowett]

Won't this result in us failing to launch on terminals who mangled their responses (like, ghostty)?

Should we display a message during startup that later gets destroyed? can we just use 10 seconds or something absurdly high?

[j4james]

Won't this result in us failing to launch on terminals who mangled their responses (like, ghostty)?

I was going to say this shouldn't be a problem, because everyone should at least respond to the DA1 query (and I know ghostty does). But I've just now seen the issue that was raised here a couple weeks ago where the OSC 4 queries somehow managed to confuse their parser so much that it wasn't responding to anything. That's a new one for me.

can we just use 10 seconds or something absurdly high?

Yeah. That was what I was thinking as well. If the terminal is broken, then it's their own fault that they have a delay at startup. But you don't want the app to break on a working terminal, just because it happens to be on a slow connection.