rethinking syscall tapset

[Frank Ch. Eigler]

hunt wrote:

> [...]
> 5. The real problem will be if we still run into compiler-related
> problems where some variables are unavailable.

Making a test run in unoptimized mode will help find (but not fix)
those.

> I'll try to work out the documentation format.  Then maybe we can
> split the syscalls and each take half?

It may be sufficient to document
- the *general* convention of the special variables ("tracestr" etc.),
  struct field explosions, type conversions
- any *differences* between the standard system call man page and
  the tapset implementation

In other words, focus on the novel, and don't bother write a new form
of the fact that "read" takes an integer "fd" argument, etc.

- FChE