feat(simulator): virtual drone fleet data generator

One process per drone: seeded patrol kinematics, sensor streams at
realistic rates, detection events, hot current/ blocks sealed into
Hive-partitioned Parquet, and the compact UDP state broadcast with
peer RSSI derivation. Scales via docker compose --scale.

simulator/virtual_drone/sensors.py

@@ -0,0 +1,80 @@
+"""Sensor row generators. Pure functions: (pose, rng, ts) -> row dict.
+
+Every row carries int64 epoch nanoseconds. Values are emitted at full float
+precision - quantization happens only in the broadcast payload, never in
+stored telemetry.
+"""
+
+from __future__ import annotations
+
+import math
+import random
+from typing import Any
+
+from .flight import Pose
+
+Row = dict[str, Any]
+
+
+def imu_row(pose: Pose, rng: random.Random, ts_ns: int) -> Row:
+    return {
+        "ts_ns": ts_ns,
+        "accel_x": rng.gauss(0.0, 0.35),
+        "accel_y": rng.gauss(0.0, 0.35),
+        "accel_z": rng.gauss(-9.81, 0.25),
+        "gyro_x": math.radians(rng.gauss(pose.roll, 0.8)),
+        "gyro_y": math.radians(rng.gauss(pose.pitch, 0.8)),
+        "gyro_z": math.radians(rng.gauss(0.0, 0.5)),
+    }
+
+
+def baro_row(pose: Pose, rng: random.Random, ts_ns: int) -> Row:
+    pressure = 101325.0 * math.exp(-pose.z / 8434.0) + rng.gauss(0.0, 4.0)
+    return {"ts_ns": ts_ns, "pressure_pa": pressure, "alt_est_m": pose.z + rng.gauss(0.0, 0.3)}
+
+
+def temp_row(_pose: Pose, rng: random.Random, ts_ns: int) -> Row:
+    return {
+        "ts_ns": ts_ns,
+        "cpu_c": 55.0 + rng.gauss(0.0, 3.0),
+        "gpu_c": 62.0 + rng.gauss(0.0, 4.0),
+        "ambient_c": 24.0 + rng.gauss(0.0, 0.5),
+    }
+
+
+def battery_row(_pose: Pose, rng: random.Random, ts_ns: int, elapsed_s: float, duration_s: float) -> Row:
+    level = max(0.0, 100.0 - 80.0 * elapsed_s / max(duration_s, 1.0) + rng.gauss(0.0, 0.2))
+    return {"ts_ns": ts_ns, "level_pct": level, "voltage_v": 22.2 * (0.85 + 0.15 * level / 100.0)}
+
+
+DETECTION_CLASSES = ("vehicle", "person", "animal", "structure", "unknown")
+
+
+def detection_row(pose: Pose, rng: random.Random, ts_ns: int) -> Row:
+    return {
+        "ts_ns": ts_ns,
+        "cls": rng.choice(DETECTION_CLASSES),
+        "confidence": round(rng.uniform(0.42, 0.99), 3),
+        "obj_x": pose.x + rng.uniform(-40.0, 40.0),
+        "obj_y": pose.y + rng.uniform(-40.0, 40.0),
+        "obj_z": rng.uniform(0.0, 5.0),
+        "ego_x": pose.x,
+        "ego_y": pose.y,
+        "ego_z": pose.z,
+    }
+
+
+def state_row(pose: Pose, ts_ns: int) -> Row:
+    return {
+        "ts_ns": ts_ns,
+        "pos_x": pose.x,
+        "pos_y": pose.y,
+        "pos_z": pose.z,
+        "roll": pose.roll,
+        "pitch": pose.pitch,
+        "yaw": pose.yaw,
+        "vel_x": pose.vx,
+        "vel_y": pose.vy,
+        "vel_z": pose.vz,
+        "frame_ref": 0,
+    }