A slightly different approach. In terms of syscalls it probably doesn't
matter at all, but it's a simpler, less hacky and maybe more efficient
way to work around PATH_MAX limitations than constantly chdir()'ing for
each directory.  Should also fix the chdir("..") fail case in MacOS
firmlink scenarios, though I haven't a clue if MacOS implements all
these openat()-related calls in the first place.

Downsides:
- Less portable. Everything is in POSIX, but it's more recent than what
  we used to use.
- Requires more open file descriptions, might hit the limit for deeply
  nested directories.

Haven't done much testing yet, but there's a bunch of TODO's:
- Make refreshing work again (not sure what's going wrong there)
- See if we can report readdir() etc errors again?
- Make firmlink detection work again?
- See how portable this really is.
- Also update file deletion code

Fixes #171

Use 'max(10, column_size / 7)' instead of a fixed size of 10

What a mess.

https://code.blicky.net/yorhel/ncdu/issues/153#issuecomment-764

Avoid undeclared identifiers when compiling with older kernel headers.

So we get around the PATH_MAX limitation. Also a tiny bit more
efficient, I hope.

Silly one-character typo that causes directory apparent sizes to be very
off in some scenarios.

Reported & patched by Andrew Neitsch.

Fixes #150

(cherry picked from commit a076ac714aa6fb24e0e9eb227372df84a4b6000b)

(cherry picked from commit ef7b4e5c28739c26b3a120296f9dc2ddaf2d4375)

(cherry picked from commit 9f28920a64c155957b83324cd3a3cf0fda6020c9)

(cherry picked from commit ad5b7fce74a3b7b0ed726620c81ea552c643cdad)

(cherry picked from commit 2faefc3b2432409f3ccca0b6ab53a228d5b9b00d)

(cherry picked from commit 32b77d0064d3e5f52e9204ddeffc29bb21a67753)

(cherry picked from commit ce7036d249b6f05633c343ab95d88552aab85402)

(cherry picked from commit 9801f46ece0ca2525f02d71464efc42296dddcb5)

(cherry picked from commit e4e8ebd9e0037812436a1588809deb23e0f3751a)

* Use AS_HELP_STRING instead of deprecated AC_HELP_STRING
* Use AC_OUTPUT without arguments
* Enclose AC_INIT argument in brackets
* Add automake option std-options

(cherry picked from commit 53a33e1db2c9a58b867f7db26c5f476c3ee4b00b)

(cherry picked from commit fd75bd0c22a0ab6935ac5332999ebe77c6d8a1c8)

Fixes #141

Reduces memory by a tiny bit. Arguably we never needed tombstones
because entries are never removed, so there shouldn't be any performance
hit there. We don't even need a 'used' flag either, considering that can
be represented by a NULL value, but I'm not really up for
implementing/modifying my own hash table.

ref: https://attractivechaos.wordpress.com/2019/12/28/deletion-from-hash-tables-without-tombstones/

They were defined (rather than just declared) in .h files, when they
really should be defined in at most a single .c file.

Fixes #140.

I assumed that ./configure would fail if it can't find a suitable
compiler for the given --host, but it just falls back to the host gcc.

Fixes #138

Fixes #135

Broken in 77aca35f
Fixes #134