Update the pin to pymeos-cffi >=1.4.0 and version to 1.4.0-alpha-1.
Rename the call sites that switched from MEOS 1.3's tpoint family to
the 1.4 tgeo/tspatial families:

- temporal-output renames: geoset_as_ewkt/as_text/out -> spatialset_*,
  geoset_round -> set_round, geoset_srid -> spatialset_srid,
  pgis_geometry_in/geography_in -> geom_in/geog_in
- tpoint_to_stbox -> tspatial_to_stbox
- tpoint_at_stbox/minus_stbox -> tgeo_at_stbox/minus_stbox
- tpoint_at_geom_time(t, gs, None, None) -> tpoint_at_geom(t, gs)
- tpoint_minus_geom_time(t, gs, None, None) -> tpoint_minus_geom(t, gs)
- tcontains/tdisjoint/tdwithin/tintersects/ttouches_tpoint_* ->
  *_tgeo_*, with the restr/atvalue trailing args dropped
- edisjoint/edwithin/eintersects/etouches_tpoint_* -> *_tgeo_*
- nad/nai/distance_tpoint_* -> *_tgeo_*
- tpoint_extent_transfn -> tspatial_extent_transfn
- spanset_make: dropped normalize/order args (1.4 signature is 1-arg)
- STBox.expand_space: tpoint_expand_space/geo_expand_space were
  removed in 1.4; rebuild as (geo|tspatial)_to_stbox + stbox_expand_space

Pytest collects 4480 tests on this branch. Test execution still
trips on functions that need shape metadata in MEOS-API (e.g.
stbox_quad_split's count output param); follow-up after MEOS-API
adds those entries.

temporal_to_tsequence / temporal_to_tsequenceset now take the
InterpolationType enum directly; PyMEOS was passing the legacy string
form. Pass the IntEnum value (which is already what TInterpolation
holds underneath).

temporal_append_tinstant gained an interp argument in 1.4. It only
matters when the receiver is a TInstant being promoted to a TSequence;
default to STEP because LINEAR is rejected by MEOS for discrete-only
base types like bool/text/integer.

Cherry-picks the equivalent of bump/meos-1.3 commits 52e8237 +
d8f9c02 + 22110ef. PyMEOS 1.4 needs the same workflow shape: apt
postgresql-server-dev-16, head_ref-first PyMEOS-CFFI checkout, and
MEOS branch derived from the package version (so 1.4 picks up MEOS
master / future stable-1.4 automatically).

MEOS 1.4 added an explicit `interp` argument to
temporal_append_tinstant (1.3 inferred it internally from the
temporal type's continuity). The bump stopgapped it with a hardcoded
InterpolationType.STEP, so appending an instant to a TInstant of a
continuous type (TFloat / TGeomPoint / TGeogPoint / …) promoted it to
a STEP sequence instead of LINEAR.

Restore the 1.3 semantics: derive the argument from the base type's
continuity via a new Temporal._continuous class attribute (True by
default; False on the step-only TInt / TBool / TText, for which MEOS
rejects LINEAR). The argument is consulted by MEOS only on the
instant->sequence promotion and ignored otherwise.

Suite: 4482 passed (was 4479); all 24 append_instant cases green
across every temporal type, no regressions.

Companion to the MobilityDB 1.4 surface (geodetic-consistent temporal
spatial relationships and set-theoretic bounding-box adjacency). Three
test-data alignments, each verified against a locally built MEOS 1.4:

* test_temporal_withindist[Discrete Sequence Geo]: the expected was the
  planar answer ({True, True}: sqrt(2) <= 2), produced when geodetic
  dwithin was wrongly computed with the planar kernel. The geodesic
  distance from POINT(1 1) to POINT(2 2) is ~156 876 m, far beyond the
  2 m threshold, so the correct geodetic result is
  {True@2019-09-01, False@2019-09-02}.

* test_temporal_touches -> test_temporal_touches_geodetic_not_supported:
  ttouches is the genuine DE-9IM topological predicate, not tdwithin(0);
  MEOS does not define it for geodetic coordinates. Assert the
  documented, intentional MeosInvalidArgValueError, mirroring the
  in-file test_speed_without_linear_interpolation_throws precedent.

* TestTFloatTopologicalFunctions::test_is_adjacent: MEOS 1.4 computes
  bounding-box adjacency set-theoretically -- two boxes are adjacent
  when their closures meet at a shared boundary value in any dimension,
  independent of the inclusive/exclusive flag. The four [v, v] float
  value-span cases that share their boundary value are therefore now
  adjacent (True); the previous False answers reflected the pre-1.4
  rule that also required exactly one inclusive bound. Integer value
  spans canonicalise to half-open [v, v+1), so the analogous tint cases
  are unaffected and keep returning False.

Make spatial_ref_sys.csv reachable at the MEOS default path in CI. MEOS
reads the CSV from the hard-coded default /usr/local/share on a source
install; on runners whose install prefix differs (macOS arm64 ->
/opt/homebrew) the file is installed elsewhere, so geography SRID
validation failed at collection with "Only lon/lat coordinate systems
are supported in geography". Mirror the file to the default path when
the prefix is not /usr/local.

Also reflow tpoint.py, temporal.py, tfloat_test.py and datespan.py with
Black so the lint job passes across the whole repository.

Verified: full PyMEOS suite 4492 passed against MEOS 1.4.