fix: don't crash exec_all_mem_mutations on an empty batch

[venimus]

What

exec_all_mem_mutations/2 crashes with MatchError: no match of right hand side value: nil when the batch has zero blocks: List.last(blocks_mutations)
returns nil on an empty list, and matching {{height, _mbi}, _block, _mutations} = nil raises.

Live incident (confirmed, not theoretical)

Hit this in production on mainnet shortly after
today's #2202/#2203 merges: crashed 48 times in a ~100ms burst, exhausted
AeMdw.Sync.Server's @max_restarts, and the sync GenStateMachine went idle
(mdw_gens_per_minute: 0.0, no forward progress) — matching the previously
documented "crashes N times then sleeps up to 1h" behavior. Confirmed via the
actual stacktrace in the deployed node's logs:

** (MatchError) no match of right hand side value: nil
    (ae_mdw 1.108.0) lib/ae_mdw/sync/server.ex:494: AeMdw.Sync.Server.exec_all_mem_mutations/2
    (ae_mdw 1.108.0) lib/ae_mdw/sync/server.ex:340: anonymous fn/3 in AeMdw.Sync.Server.spawn_mem_sync_from/3

Fix

Treat an empty batch as a legitimate no-op (return state unchanged) instead
of crashing — this can happen when a resume/extension check computes a
from_height that turns out to have nothing new to sync past the
already-synced height.

Verification

• Live-patched mainnet mdw1 directly (hot-load, same procedure as earlier
  today) to stop the active crash loop immediately: confirmed zero new
  crashes post-patch, mdw_gens_per_minute jumped from 0.0 to 512+ and
  height caught up rapidly.
• mix format --check-formatted && mix credo — clean.
• mix dialyzer --format github — passed successfully.
• mix test test/ae_mdw/sync/ test/ae_mdw/db/sync/ — 55/55 passing.