…ch_path`.

Instead of maybe storing its own sysroot and maybe deferring to the one
in `Session::opts`, just clone the latter when necessary so one is
always directly available. This removes the need for the getter.

It's more idiomatic, makes the code shorter, and will help with the next
commit.

`FileSearch::search()` traverses one or more directories. For each
directory it generates a `Vec<PathBuf>` containing one element per file
in that directory.

In some benchmarks this occurs enough that the allocations done for the
`PathBuf`s are significant, and in practice a small number of
directories are being traversed over and over again. For example, when
compiling the `tokio-webpush-simple` benchmark, two directories are
traversed 58 times each. Each of these directories have more than 100
files.

This commit changes things so that all the `Vec<PathBuf>`s that will be
needed by a `Session` are precomputed when that `Session` is created;
they are stored in `SearchPath`. `FileSearch` gets a reference to the
necessary `SearchPath`s. This reduces instruction counts on several
benchmarks by 1--5%.

The commit also removes the barely-used `visited_dirs` hash in
`for_each_lib_searchPath`. It only detects if `tlib_path` is the same as
one of the previously seen paths, which is unlikely.

Returning an iterator leads to nicer code all around.

In an attempt to avoid "thread '<unnamed>' panicked at 'failed to
acquire jobserver token: Bad file descriptor" errors.