Business plan draft 0.3 in two versions: Plan A ~180M and Plan B ~225M THB

- One build makes both documents: plan_variant.py switches the
  plan-specific text (pick / is_a / label); build_plan.py writes
  CNX-Vinyl-Business-Plan-A and -B (docx), export_pdf.py exports both PDFs
  with the table of contents filled in. Draft 0.2 (single launch-plant
  plan with contingency) is removed; it stays in the history.
- Plan A: the launch plant with every machine and six cost levers, ~177M
  THB with the ramp-up reserve, no contingency. Plan B: the second Duplex
  at launch, five presses, ~224M THB on the same basis. Each version reads
  as a complete plan with one table comparing it with the other; the
  216M investor-question text is gone, Pheenix is priced at OMS's
  reference plus 20 %.
- Calculator renders in both documents, each from its own preset: the
  plant at a glance (Figure 4), the budget check (Figure 5) and ten years
  of cash (Figure 6), in docs/renders/calculator/.
- Calculator budget card: the verdict and third KPI now name the tightest
  budget the midpoint fits (Plan A: 3.3M to spare under 180M; Plan B:
  0.8M under 225M) instead of always 225M.
- Fixes: CD replication cross-references point to section 4.10; the
  blended price in section 3.3 and roadmap item 1 is 3.76 to 5.27 USD at
  60 % specials (3.50 to 4.60 at a standard-heavy mix).

docs/roadmap.md

@@ -8,8 +8,11 @@ Rough durations are estimates.
 
 These three run in parallel and gate everything downstream:
 
-1. **Market validation**: test the 3.50 to 4.60 USD blended pressing price
-   and regional demand with Asian labels and distributors, starting at
+1. **Market validation**: test the blended pressing price (3.76 to 5.27
+   USD at the planned 60 % specials; 3.50 to 4.60 at a standard-heavy mix)
+   and regional demand, which decides between Plan A (~0.74M records a year)
+   and Plan B (~1.25M) in [total investment](finance/total-investment.md#two-plans-a-at-180m-thb-and-b-at-225m-thb-october-2026),
+   with Asian labels and distributors, starting at
    Making Vinyl in Haarlem (Alexander and Simona). This is the biggest
    assumption in the [revenue model](finance/revenue-model.md); the
    financing case waits on it. **Partially started**: a first buy-side