This makes the build tree more organized, which is good for adding new
stuff.  Now, it's useful for example for adding manual pages in man3/,
but it may be useful in the future for example for extending the build
system to run linters (e.g., clang-tidy(1), Clang analyzer, ...) on the
C source code.

Previously, the build tree was quite flat, and looked like this (after
`./configure && make`):

    $ tree -I src build
    build
    ├── Makefile
    ├── autoconf.data
    ├── autoconf.err
    ├── echo
    ├── libnxt.a
    ├── nxt_auto_config.h
    ├── nxt_version.h
    ├── unitd
    └── unitd.8

    1 directory, 9 files

And after this patch, it looks like this:

    $ tree -I src build
    build
    ├── Makefile
    ├── autoconf.data
    ├── autoconf.err
    ├── bin
    │   └── echo
    ├── include
    │   ├── nxt_auto_config.h
    │   └── nxt_version.h
    ├── lib
    │   ├── libnxt.a
    │   └── unit
    │       └── modules
    ├── sbin
    │   └── unitd
    ├── share
    │   └── man
    │       └── man8
    │           └── unitd.8
    └── var
        ├── lib
        │   └── unit
        ├── log
        │   └── unit
        └── run
            └── unit

    17 directories, 9 files

It also solves one issue introduced in
5a37171f733f ("Added default values for pathnames.").  Before that
commit, it was possible to run unitd from the build system
(`./build/unitd`).  Now, since it expects files in a very specific
location, that has been broken.  By having a directory structure that
mirrors the installation, it's possible to trick it to believe it's
installed, and run it from there:

    $ ./configure --prefix=./build
    $ make
    $ ./build/sbin/unitd

Fixes: 5a37171f733f ("Added default values for pathnames.")
Reported-by: Liam Crilly <liam@nginx.com>
Reviewed-by: Konstantin Pavlov <thresh@nginx.com>
Reviewed-by: Andrew Clayton <a.clayton@nginx.com>
Cc: Andrei Zeliankou <zelenkov@nginx.com>
Cc: Zhidao Hong <z.hong@f5.com>
Signed-off-by: Alejandro Colomar <alx@nginx.com>

The patch removes the "files" section from package.json to avoid future issues
with missing files.  For package testing purposes, 'npm pack' is used instead
of plain 'tar' to simulate packaging more accurately.

This makes the "sed" instruction simpler and more portable, as the previous
variant didn't work well on BSD systems due to the "\s" metacharacter.

Thanks to Sergey A. Osokin <osa@FreeBSD.org.ru> for spotting this issue.

Also, this should prevent accidentally creating a version 1.0.0 package.

Globally installed modules require a globally installed libunit.
The "binding_pub.gyp" file is the correct version, otherwise linked module
may have unresolved symbols because libunit is not linked.

While it looks nicer without zero 3-rd version number, this should
improve interoperability.  Version string can be parsed or used for
sorting.  And it is easier to handle and less confusing when there
is constant number of version parts.

Moreover, NPM also expects version format with 3 parts.
So ".0" has already been used in Node.js module version.

This also eliminates expressions that incompatible with BSD make, thus fixing
installation of Node.js module on FreeBSD (broken by dace60fc4926).

By default "npm install" switches to non-privileged user to run package scripts
if it is invoked by root.  As a result it may prevent node-gyp from writing to
package directory and break installation of the module.

To disable this switching the --unsafe-perm flag is added.