docs: explain why NUTS starts take one walker per lineage

   harmless when NUTS follows, because only the lowest-energy walkers are used.

   The re-seeding is needed also when NUTS follows: the spread of the walkers

   decides between NUTS and ensemble production (step 3) and sets the NUTS mass

   matrix, and stuck walkers would distort both. A re-seeded walker, however,

   continues the history of the walker it was copied from, so NUTS never starts

   two chains from walkers of the same lineage (step 4).

   matrix, and stuck walkers would distort both. A re-seeded walker,

   however, is a copy that continues the history of its donor, so NUTS never

   starts two chains from the same lineage (step 4; "NUTS mechanics" below).

3. **Phase 2, energy-spread test.** Compares the spread of walker potential

   energies with `chain_energy_threshold × n_free_params`.

4. **Phase 3a, unimodal → NUTS.** If the spread is ≤ threshold, NUTS chains

  chains that did not share their starting point make R-hat meaningful: R-hat

  compares the spread between chains with the spread within them, and chains

  started from one point agree with each other for a while even when they have

  not converged. For the same reason the starts are taken at most one per

  lineage: a walker re-seeded from another during the warmup and the walker it

  was copied from count as one, and the next lowest energy is taken instead

  (only if there are fewer lineages than chains do two chains share one). Until

  October 2026 all chains started from the single best walker, and until

  2026-10-05 a re-seeded walker and its donor could both be chosen. R-hat above 1.01 is flagged in the diagnostics table (`*`, and `**`

  not converged, so R-hat then looks better than it should. A walker restarted

  from another during the warmup (step 2) is a copy: it continues the history of

  its donor and, for a while, moves like it. Chains started from a copy and its

  donor would therefore agree for the same reason, so PyAETNA records the

  *lineage* of every walker (the original walker whose history it continues) and

  takes the starts at most one per lineage, the next lowest energy replacing a

  duplicate. This keeps the restart of stuck walkers, which the route decision

  and the mass matrix need, while the chains start independently. Only if there

  are fewer lineages than chains do two chains share one. Until October 2026 all

  chains started from the single best walker, and until 2026-10-05 a restarted

  walker and its donor could both be chosen. R-hat above 1.01 is flagged in the diagnostics table (`*`, and `**`

  above 1.1).

- **Frozen inverse mass matrix**: a diagonal inverse mass matrix from the

  variance of the walker positions (`adapt_mass_matrix=False`). Only the step