[JasMehta08]

What

Adds docs/research/workload-model.md: the one workload we price on AWS, Azure and on-prem. Every parameter has a source ID or is labelled as an assumption.

Closes #15

Why

#23, #24, #25 and #26 need the same precisely stated workload. This fixes it: data volume, growth, block size, requests, egress, hot/cold split, durability and horizon.

Decisions to confirm in review

• Baseline: 2016 at 500 PB [B1].
• Horizon: 5 years.
• Scale sweep: 0.1, 1, 10, 100 and 500 PB, so [Cost] Compare AWS vs Azure vs on-prem and find the break-even point #26 can find the break-even point.
• Assumptions: 2 MB average block, egress = 2 x ingest, no deletes.
• Cloud default: single region, with a second-region scenario for [Cost] Compare AWS vs Azure vs on-prem and find the break-even point #26.

How to check it

• Section 2: every row has a source ID or "Assumption".
• Sections 3 and 4: derived tables. Spot-check year 1: 100 PB / 2 MB = 50 billion PUTs.
• Section 6: sanity checks against the S-1 [F1] and Dropbox's own figures.

Checklist

• cargo fmt, cargo clippy and cargo test pass locally (code PRs)
• Tests added or updated for the change (code PRs)
• New sources are added to docs/sources/README.md (research PRs)
• Design decisions are recorded in the Design Decision Log
• Docs and README are updated if behaviour or setup changed