Challenge 2: Decide — Bucket or Not? — Possible Solution
====================================================================

YES, this should use the bucket pattern.

WHY: one reading every 5 seconds is 12 readings per minute, 720 per
hour — for a single workout (typically 20-60 minutes), that's easily
several hundred to over a thousand tiny documents, each holding
essentially nothing but a timestamp and a number. That's precisely the
"many tiny, frequent documents" signal the chapter names: a large
document count relative to how little each document actually holds,
plus an index entry paid for every single one of them.

PROPOSED BOUNDARY: bucket per-workout, not per-minute.

A workout is a natural, bounded unit — it has a clear start and end,
it's the unit the app actually queries by ("show me this workout's
heart-rate graph"), and even at one reading every 5 seconds, a full
hour-long workout is only 720 readings — comfortably small for a
single document, nowhere near MongoDB's 16MB limit.

Per-minute would be technically valid but adds unnecessary complexity:
the app would need to query and stitch together many bucket documents
just to render one workout's graph, when the natural query boundary
(one workout) already gives a bucket size that's small enough to keep
as one document.