feat: add memory-aware parallel merge scheduling and CLI flags

Introduces a memory-aware scheduling strategy for parallel partition merging that replaces unbounded concurrency with a First-Fit Decreasing approach gated by a thread-safe `MemoryBudget` semaphore. An adaptive expansion factor, seeded by a sequential pilot run, dynamically caps concurrent workers to prevent hashbrown OOMs. Adds a `--budget-fraction` CLI flag to configure RAM allocation, enhances the CLI to accept multiple indexes, and introduces comprehensive partition diagnostics including memory utilization tracking, concurrency metrics, and statistical summaries with ASCII histograms. Updates documentation and navigation accordingly.

src/obikpartitionner/src/merge_layer.rs

@@ -166,10 +166,10 @@ impl KmerPartition {
         n_dst_genomes: usize,
         block_bits: u8,
         evidence: &IndexMode,
-    ) -> SKResult<()> {
+    ) -> SKResult<usize> {
         let dst_index_dir = self.part_dir(i).join(INDEX_SUBDIR);
         if !dst_index_dir.exists() {
-            return Ok(());
+            return Ok(0);
         }
 
         load_meta(&dst_index_dir)?; // ensure meta.json exists before LayeredMap::open
@@ -381,6 +381,6 @@ impl KmerPartition {
             part_meta.save(&dst_index_dir).map_err(olm_to_sk)?;
         }
 
-        Ok(())
+        Ok(n_new)
     }
 }