fix: add area column to CSVSink BASE_HEADER

[Zeesejo]

Problem

CSVSink writes bounding-box coordinates, confidence, class_id, and tracker_id to CSV but silently omits the area of each detection, even though Detections.area is a standard computed property. Users trying to log area either get an empty column or an AttributeError when passing it via custom_data.

Reported in #1397.

---

Root Cause

BASE_HEADER (the static list of always-written columns) never included "area":

# before
BASE_HEADER = [
    "x_min", "y_min", "x_max", "y_max",
    "class_id", "confidence", "tracker_id",
    # ← area missing
]

parse_detection_data() built each row from the xyxy array + optional fields, but never called detections.area.

---

Fix

Two small, self-contained changes to csv_sink.py:

1. BASE_HEADER — add "area"

BASE_HEADER = [
    "x_min", "y_min", "x_max", "y_max",
    "class_id", "confidence", "tracker_id",
    "area",   # ← added
]

2. parse_detection_data() — compute and include area per row

areas = detections.area          # computed once outside the loop
for i in range(len(detections.xyxy)):
    row = {
        ...,
        "area": str(areas[i]),   # ← added
    }

detections.area is computed once before the loop (not per-iteration) to avoid redundant work on large batches.

---

Verification

import supervision as sv
import numpy as np

detections = sv.Detections(
    xyxy=np.array([[10, 20, 110, 120]]),  # area = 100*100 = 10000
    confidence=np.array([0.95]),
    class_id=np.array([0]),
)

with sv.CSVSink("out.csv") as sink:
    sink.append(detections)

# out.csv now contains an 'area' column with value 10000.0

Fixes #1397

[Copilot]

Pull request overview

This PR fixes CSVSink serialization to include each detection’s computed area in the emitted CSV, addressing the gap reported in issue #1397.

Changes:

• Add "area" to CSVSink’s BASE_HEADER so it’s always written.
• Compute detections.area once and include the per-detection value in each serialized row.

[Copilot]

Adding area to BASE_HEADER can produce duplicate area columns when callers still pass custom_data={'area': ...} (or if detections.data contains an area key). parse_field_names() currently appends all dynamic keys without excluding base fields, so the header can contain duplicate names and the resulting CSV becomes ambiguous. Consider filtering dynamic_header to remove any keys already in BASE_HEADER (and/or treating base-field overrides as an error).

Suggested change

	]
	]
	BASE_HEADER_SET = set(BASE_HEADER)

[Copilot]

This change updates the CSV schema (new area column) and will require updating existing CSVSink tests/fixtures that assert the header and row layouts (e.g., tests/detection/test_csv.py expected headers/rows). Please add/adjust test expectations to include area in the correct position and with the computed values so CI stays green.