install: Simplify path construction.

There's no need to copy the path twice to split it into base and dir.
We simply call `basename()` first, then handle the two trivial cases in
which it isn't safe to call `dirname()`.

While here, add an early check that the destination is not an empty
string. This would always fail eventually, so it may as well fail
right away. Also add a test case for this shortcut.

MFC after: 1 week
Sponsored by: Klara, Inc.
Reviewed by: markj
Differential Revision: https://reviews.freebsd.org/D44743

install: handle -m +X more accurately

As described by chmod(1), +X in the mode may be used to optionally set
the +x bit if the file is a directory if any of the execute/search bits
are set in the original mode. The latter is not applicable because we
assume -m is a fresh mask, but a functional +X could be useful in the
former case if we're passing along a common INSTALL_MODE that's designed
to install either 0644 or 0755 depending simply on whether it's a
directory or not.

Reviewed by: des
Sponsored by: Klara, Inc.
Differential Revision: https://reviews.freebsd.org/D42273

Remove $FreeBSD$: one-line sh pattern

Remove /^\s*#[#!]?\s*\$FreeBSD\$.*$\n/

xinstall: fix dounpriv logic, add tests

Sponsored by: Klara, Inc.
MFC after: 1 week

Fix the tests for install(1): add support for STRIPBIN's -o option.

Reported by: lwhsu
MFC after: 1 week

install(1): Fix relative path calculation with partial common dest/src

For example, from the referenced PR [1]:

$ mkdir /tmp/lib/ /tmp/libexec
$ touch /tmp/lib/foo.so
$ install -lrs /tmp/lib/foo.so /tmp/libexec/

The common path identification bits terminate src at /tmp/lib/ and the
destination at /tmp/libe. The subsequent backtracking is then incorrect, as
it traverses the destination and backtraces exactly one level while eating
the 'libexec' because it was previously (falsely) identified as common with
'lib'.

The obvious fix would be to make sure we've actually terminated just after
directory separators and rewind a character if we haven't. In the above
example, we would end up rewinding to /tmp/ and subsequently doing the right
thing.

Test case added.

PR: 235330 [1]
MFC after: 1 week

Restore some of the error message text accidentally removed in r315098

"unexpected symlink contents" is more pedantically correct than
"unexpected symlink".

MFC after: 1 week
X-MFC with: r315098
Sponsored by: Dell EMC Isilon

Add 3 more testcases demonstrating how install -l sr works

The additional testcases use absolute paths for sources and targets,
as the other testcase which tested `-l sr` used flat relative paths in
the same directory.

Please note that these testcases do not test `-l a` -- that's already
addressed in the battery of tests.

MFC after: 1 week
Sponsored by: Dell EMC Isilon

Clarify src vs dest path mismatch in :symbolic_link_{absolute,relative}_body

Unfortunately kyua does not omit the path mismatch on failure, so it must be coded
into the error message.

Cache the values, run the test(1) call, then print out the values in an atf_fail
call to emit the required diagnostics to debug why things are failing.

MFC after: 1 week
Sponsored by: Dell EMC Isilon

install: When preserving timestamps, also copy the nanoseconds part.

Now that we have utimensat in -legacy, install(1) can use it.

This is a revert of r299942 which is itself a revert of r299850.

install: Revert utimensat usage (r299850).

This should fix the build on older stable/10, since install is a bootstrap
tool.

Pending a decision how to fix this properly, revert utimensat usage. Copies
with the -p option will again appear older than the original almost always,
but -p is not commonly used.

install: When preserving timestamps, also copy the nanoseconds part.

install: Add some tests.