Compile the module against the installed MEOS umbrella header

[estebanzimanyi]

The C wrappers compile against an installed MEOS prefix rather than the MobilityDB source tree, which is the header MEOS actually publishes.

MEOS assembles the installed <meos.h>: it splices its PostgreSQL-compat definitions into it under #ifndef POSTGRES_H, and that block supplies Datum, DateADT, TimestampTz, Interval and the de-prefixed base I/O — interval_in, date_in, timestamptz_in and siblings, named without a meos_ prefix so they do not collide with a real PostgreSQL. None of it appears in the source copy, so calls to those functions are implicitly declared and C99 rejects them.

• the build installs MEOS and compiles against that prefix;
• the preamble stops including the vendored <postgres.h>, which defines POSTGRES_H first and makes <meos.h> skip the entire spliced block, leaving even DateADT undefined, and takes <string.h> directly rather than through it;
• the Datum accessors the spliced block leaves out are defined in the preamble, keyed on the pointer width exactly as PostgreSQL keys USE_FLOAT8_BYVAL on SIZEOF_VOID_P >= 8.

That last point decides behaviour, not just compilation. With -sMEMORY64=1 a Datum is 8 bytes, so a TimestampTz or an int64 span bound rides inside the Datum rather than behind a pointer. Reading one through a pointer returns whatever the value addresses, which makes BigIntSpan [1,5) and [6,10) report as adjacent. Keying on the width keeps both models correct.

docker build --target wasm exits 0 with no errors.

Under Node 24 the suite runs and fails 13 tests, the same 13 as a module built from the commit before this branch. Node 22 raises CompileError: invalid table elements limits flags before any test runs, since V8 12.4 does not accept the 64-bit tables -sMEMORY64=1 emits and V8 13.6 does.