fix(blocks): clamp tail chunk to reorg buffer max instead of dropping it

[AuburyEssentian]

Problem

Fixes #193.

When --reorg-buffer N is set, apply_reorg_buffer used filter_map to drop any chunk whose max_value() exceeded latest_block - N. This silently discarded the partial tail chunk — e.g. if the last full chunk boundary was block 20,029,999 but the reorg-safe ceiling was 20,030,779, the 780-block tail chunk Range(20030000, 20030787) was dropped entirely rather than trimmed to Range(20030000, 20030779).

The result: cryo with --reorg-buffer always rounds the max collected block down to the last full chunk boundary, losing up to chunk_size - 1 blocks.

Fix

Replace the filter_map predicate with a clamp_chunk_to_max helper that:

• For Range(start, end) chunks: returns Some(Range(start, end.min(max_allowed))) if start <= max_allowed, else None.
• For Numbers(ns) chunks: filters out block numbers > max_allowed, returning None only if the entire list is beyond the buffer.

This preserves every safe block in the tail chunk rather than discarding it.

Before / After

# Before (reorg-buffer 8, latest = 20030787)
blocks: n=30,000 min=20,000,000 max=20,029,999  ← tail chunk silently dropped

# After
blocks: n=30,779 min=20,000,000 max=20,030,779  ← tail chunk clamped correctly