Reject zero-area shapes with meaningful errors

[progys]

Problem

Two robustness issues with zero/degenerate shapes:

1. Zero-area triangles were silently accepted and stored (e.g. `triangle 0 0 0 0 1 1`). `Triangle.inShape` then divided by `2 * area` = 0, producing `Infinity`/`NaN` and silently returning false for every point.
2. `donut 0 0 0 2` (zero inner radius) failed with a misleading message — the error surfaced from the inner `Circle` constructor ("Circle radius should be positive number") instead of naming the donut.

Both violate requirement #4 ("meaningful error message and continue the execution").

Fix

• `Triangle` constructor now rejects collinear/degenerate vertices (`area == 0`) — the divide-by-zero in `inShape` is unreachable by construction.
• `Donut` constructor validates inner/outer radii positivity with clear messages before constructing its circles.

Tests (21 total, up from 11)

• `TriangleTest`: degenerate + collinear rejection; inside / outside / on-boundary containment.
• `DonutTest`: zero/negative/equal-radius rejection; point in ring / hole / outside outer circle.
• End-to-end: every invalid zero-radius input prints a meaningful message and execution continues.

[cursor]

Bugbot is not enabled for your account, so this pull request was not reviewed.

Enable Bugbot in the Cursor dashboard to get automatic reviews on future PRs.