fix: correct N2 learning opcodes (18→14) in comparison table and data

Both N1 and N2 have 14-opcode learning ISA. The N2's differentiator is
programmable neuron microcode, not more learning opcodes.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

app.py

@@ -37,7 +37,7 @@ PROCESSORS = {
         "total_neurons": 131072,
         "dendrites": 4,
         "graded_spike_bits": 8,
-        "learning_opcodes": 18,
+        "learning_opcodes": 14,
         "max_axon_delay": 63,
         "parity": "Intel Loihi 2",
         "neuron_models": ["LIF", "CUBA", "ALIF", "Izhikevich", "Custom"],
@@ -46,7 +46,7 @@ PROCESSORS = {
             "CUBA, LIF, ALIF, Izhikevich built-in",
             "Custom neuron models via microcode",
             "Graded spikes (8-bit payload)",
-            "On-chip learning (18-opcode ISA)",
+            "On-chip learning (14-opcode ISA)",
             "Dendritic compartments (4 per neuron)",
             "Axonal delays (up to 63 timesteps)",
             "Multi-chip scalability",
@@ -375,7 +375,7 @@ Configure networks, validate hardware constraints, and run simulations — all i
 | **Neurons/core** | 1,024 | 1,024 |
 | **Total neurons** | 131,072 | 131,072 |
 | **Neuron models** | LIF | LIF, CUBA, ALIF, Izhikevich, Custom |
-| **Learning** | 14-opcode ISA, STDP | 18-opcode ISA, three-factor |
+| **Learning** | 14-opcode ISA, STDP | 14-opcode ISA, three-factor |
 | **Dendrites** | 4 compartments | 4 compartments |
 | **Graded spikes** | 8-bit | 8-bit |
 | **Max axon delay** | 63 timesteps | 63 timesteps |