feat(bindings): tCount(temporal) aggregate — first skiplist-backed aggregate

[estebanzimanyi]

Summary

Adds `tCount()` aggregate over all 4 temporal base types (`tint`, `tbool`, `tfloat`, `ttext`). Returns a `tint` where each instant carries the count of input temporal values valid at that instant.

```sql
-- distinct timestamps, count = 1 at each instant
SELECT tCount(t) FROM (VALUES (tint '1@2000-01-01'), (tint '2@2000-01-02')) tt(t);
-- {1@2000-01-01, 1@2000-01-02}

-- duplicate timestamp, count sums to 2
SELECT tCount(t) FROM (VALUES (tint '1@2000-01-01'), (tint '5@2000-01-01')) tt(t);
-- {2@2000-01-01}

-- overlapping sequences, count = 2 in overlap region
SELECT tCount(t) FROM (VALUES (tint '[1@2000-01-01, 1@2000-01-05]'),
(tint '[1@2000-01-03, 1@2000-01-07]')) tt(t);
-- {[1@2000-01-01, 2@2000-01-03, 2@2000-01-05], (1@2000-01-05, 1@2000-01-07]}
```

Implementation pattern (new for MobilityDuck)

This is the first skiplist-backed aggregate in MobilityDuck. The pattern established here generalises to `tand`, `tor`, `tmin`, `tmax`, `tsum`, `tavg`, `twAvg`, etc. — only the underlying `*_transfn` and merge function differ.

Component	What it does
`TCountState`	Holds a heap-allocated `SkipList *skiplist` — the skiplist itself lives outside the inline aggregate state buffer.
Registration	`AggregateFunction::UnaryAggregateDestructor` instead of `UnaryAggregate` — installs a destructor calling `TCountFunction::Destroy` → `skiplist_free`.
`Operation`	Copies the input Temporal blob (varlena) into a heap allocation and calls `temporal_tcount_transfn` — MEOS allocates a fresh skiplist on the first call and extends it in place on subsequent calls.
`Combine`	Merges two skiplists via `temporal_skiplist_splice` with a sum merge function. MEOS's `datum_sum_int32` is not exported, so we provide a local equivalent (`mduck_datum_sum_int32`).
`Finalize`	Calls `temporal_tagg_finalfn` which produces a `Temporal*` and internally calls `skiplist_free`. We null `state.skiplist` after the call so the destructor doesn't double-free.

Stacked on

• feat(bindings): aggregate function infrastructure — extent(span) for all 5 span types #47 (AggregateFunction infra)
• feat(bindings): extend extent(*) to spanset and set surfaces #48 (extent spanset/set)
• feat(bindings): extent(tnumber) → tbox + bundle extent overloads in one set #49 (extent tnumber)

Test plan

• Single-row tCount on tint/tfloat/tbool/ttext
• Multi-row distinct timestamps (count = 1 at each instant)
• Multi-row duplicate timestamps (count sums correctly)
• Overlapping sequences (count rises in overlap region)
• NULL handling: NULL ignored mid-run, all-NULL returns NULL
• No segfaults on aggregate state cleanup (destructor + Finalize ordering)

[estebanzimanyi]

Consolidated into #60 (aggregate cluster squash). The full chain of aggregate work is now reviewable as a single PR; this branch's commits are preserved in #60's history. Closing to minimize the review queue.