more flexible parsing for /proc/meminfo to take shared and reclaimable
memory into account. this matches the output with free(1).

additionally this could fix some corner cases, as the order of fields in
/proc/meminfo is not strictly defined.

slstatus:
percent 81% free 2.5 Gi total 23.4 Gi used 19.0 Gi

free(1):
               total        used        free      shared  buff/cache   available
Mem:            23Gi        19Gi       2.5Gi       1.3Gi       3.2Gi       3.6Gi

Signed-off-by: drkhsh <me@drkhsh.at>

size_t may not be enough, use 64 bit integers

On OpenBSD the backlight percentage cannot be retrieved in a simple way.
The only two solutions we are aware of for now are:

 - reading from /dev/ttyC0: which isn't possible without changing
 permissions or running slstatus as root
 - linking against xcb-xrandr: which is bloat and does not work in every
 case appearently

This reverts commit 37724ac2c3f496f4736223d0d8e5d8fecb933590 for now.

At the moment linux only, but will add support for OpenBSD as well.

It is not necessary to copy memory or anything. Just keep a pointer to
the active prefix-array and assign the length of the arrays to a
variable.

Make the code more readable by using a switch, be more strict when an
invalid base is passed to it and fix a small oversight in the bottom of
the code where the base 1024 was forgotten to generalized.

Now only one function, fmt_human, takes an additional argument "base".

This is a first step to decouple formatting from information because of
two reasons:

 1. The components should only gather and return the values by design
 2. Fine grained user control should be a focus

Scaling will be implemented in a different way in a later commit.

These functions take the raw number and a unit and automatically
print it out "scaled down" to a proper SI-prefix, for powers of 2
and 10 respectively.

Apply them to the 2-power cases and keep the 10-power for a later
commit.