Update templates, split negishi, fix williamson stereochemistry

[FabianKruger]

Changes

• 191 templates revised upstream across chan_lam, imidazole_Xketone_synthesis,
  imidazole_condensation_acid/_amine, reductive_amination_ketone, sn2_nheterocycle.
• negishi → negishi_batch (breaking). Same 44 templates; the name now says they
  need a preformed organozinc.
• negishi_insitu added (176 templates) — the one-pot variant that zincates an
  aliphatic halide in situ, so both partners are supplied as halides. Overlaps
  cross_electrophile_coupling: pairs matched by both now give the same product under
  two reaction names.
• Williamson stereochemistry fixed. Williamson is an SN2 on the halide, so that
  carbon inverts while the alkoxide carbon is retained.
  Both were previously retained. All 104 templates now mark map 4.

Verification

Williamson is checked against two reference outcomes:

butan-1-ol + (S)-2-bromooctane	(R)-2-butoxyoctane — inversion at the halide carbon
(S)-octan-2-ol + MeI	(S)-2-methoxyoctane — alkoxide carbon untouched

Negishi is checked on six known couplings across both variants, plus selectivity
(neither variant fires on two aryl halides or two alkyl halides).

Migration

reaction_list=["negishi"] → reaction_list=["negishi_batch"].