Simply tidying up.

As detailed in doc/design.lyx section 2.16 "Record Headers Are Not
Expandible", we make sure that if there is padding at the end of a record,
the first byte of padding is a zero.

As detailed in doc/design.lyx section 2.15 "Extending The Header Is
Difficult"; we add features_used and features_offered fields to the
header, so we can identify if we add new features, and then if someone
opens it who doesn't understand that feature.

This make sure that the data being handed to the function in tdb_traverse
is the correct type.

This is vitally important for forwards compatibility.

Two missed cases in b2100462 (tdb2: change API to return the error
value.)

It's now important, since it defines _GNU_SOURCE.

Recent changes shifted line numbers in tap.c, so the break is now in
the wrong place.  We should probably have an explicit function we can
breakpoint instead.

Enable CCAN_STR_DEBUG in the default flags, so our tools get checked,
and fix up the resulting errors.

The standard ctype.h variants will make you go insane if you
contemplate them for too long.

isspace() and isdigit() want an unsigned value.

tolower(), toupper() and isupper want an unsigned value.  Also, fix
lazy cost-discarding inside strrchr for extra safety.

If we detect any mention of a problematic string function, try compiling
the entire module with string debugging enabled.

In the former case, we were bitten by the fact that you don't pass a char
to isalpha() et al: you pass an int.  This means on most platforms you want
to do:

   if (isalpha((unsigned char)c)) ...

Insane?  Yes, but I assure you I'm not making this up.

Similarly, I've always wanted strstr, strchr and strrchr to return
const char * when given a const char * argument to search, to avoid
constness leak.

In both cases, the actual versions from the headers may be macros, so
we need to be very careful overriding them.  The result is that they
become out-of-line functions which is unacceptable for general
performance.

So we only activate these when CCAN_STR_DEBUG is defined.

Same thing (a BUILD_ASSERT which evaluates to zero), but there's a
strong preference for all modules to stick with their own names as
prefixes.

They're often out-of-date, and we're about to do special things to the
str module compile flags, so we don't *ever* want the default.

It doesn't add much to build times, and it eliminates a nasty source
of weird errors.

We need to flush stdout before forking.

Also add a note about a possible segfault.

We don't use it for the moment, but Brad Hards noted that assigning to
dest ten continuing was pointless, revealing this bug.

We only ever write to this variable, so can just be removed.

This is always overwritten later in the function.

obe is always overwritten later.

last is unconditionally set to zero later in this function.

lp_bits was only ever written to.

asprintf is a PITA to use, and it's not in POSIX anyway.  Provide
replacements, and also provide a nicer-to-use afmt() wrapper.

tools/ccan_depends --compile ccan/module (without a trailing slash)
results in a segfault.

(Patched forward by Rusty)

We get a point for every run or api test, rather than 1 point for all of them,
so be consistent.