Add Hebbian learning and temporal reasoning to OpenCog architecture

[Copilot]

For changes to the core ggml library (including to the CMake build system), please open a PR in https://github.com/ggml-org/llama.cpp. Doing so will make your PR more visible, better tested and more likely to be reviewed.

Description

Extends OpenCog cognitive architecture with two foundational capabilities: adaptive learning through Hebbian updates and temporal reasoning for event sequences.

Hebbian Learning

Implements "neurons that fire together, wire together" for embedding adaptation:

// Strengthen semantic connections through co-activation
ggml_opencog_hebbian_update(atomspace, dog_id, pet_id, 0.1f);

// Propagate learning through graph structure
ggml_opencog_hebbian_update_link(atomspace, inheritance_link_id, 0.15f);

// Maintain numerical stability
ggml_opencog_normalize_embedding(atomspace, atom_id);

API: 3 functions (hebbian_update, hebbian_update_link, normalize_embedding)

Temporal Reasoning

Adds time interval support with temporal relations and queries:

// Associate events with time
ggml_opencog_set_time_interval(atomspace, breakfast_id, 8000, 8500);
ggml_opencog_set_time_interval(atomspace, work_id, 9000, 17000);

// Temporal relations
bool ordered = ggml_opencog_happens_before(atomspace, breakfast_id, work_id);
bool contained = ggml_opencog_happens_during(atomspace, lunch_id, work_id);

// Query by time
auto events = ggml_opencog_get_atoms_at_time(atomspace, 10000);
auto morning = ggml_opencog_get_atoms_in_interval(atomspace, 7000, 9000);

// Sequential inference
auto inferred_tv = ggml_opencog_temporal_induction(link1_tv, link2_tv);

API: 10 functions for time intervals, queries, relations, and inference
Atom types: Added TIME_NODE, SEQUENTIAL_LINK, SIMULTANEOUS_LINK, AT_TIME_LINK

Changes

• Extended ggml_opencog_atom with time_interval field
• Increased GGML_OPENCOG_ATOM_TYPE_COUNT from 5 to 9
• Added temporal cleanup in ggml_opencog_remove_atom
• Comprehensive test coverage (11/11 tests pass)
• Example programs: opencog-hebbian.cpp, opencog-temporal.cpp

Implementation Notes

• Time intervals use int64_t millisecond timestamps
• Hebbian learning uses symmetric embedding updates with configurable rate
• Temporal induction reduces confidence along inference chains (uncertainty propagation)
• Memory managed via smart pointers; temporal data cleaned up in atom removal

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.