Fix #[track_caller] for .success(), .failure(), and more

[marcospb19]

In ouch I've long had the problem of .success() and .failure() giving me a panic that happened at $HOME/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/ops/function.rs:250:5.

use assert_cmd::Command;

fn main() {
    Command::new("false")
        .assert()
        .success();
}

Running this you get:

thread 'main' (3911719) panicked at /home/marcospb19/.rustup/toolchains/nightly-x86_64-unknown-linux-gnu/lib/rustlib/src/rust/library/core/src/ops/function.rs:250:5:
Unexpected failure.

So in this PR I fix the function wrapping that prevented #[track_caller] from working, here is what it looks like now:

thread 'main' (3914770) panicked at src/main.rs:6:10:
Unexpected failure.

I also added a bunch of #[track_caller] at places where there is a panic path, let me know if you think that's excessive.

[epage]

This seems uncommon enough and with enough context that it might not be needed

[epage]

Yeah, that is really annoying situation with combinators.