Add --detailed-read-stats flag

Summary:
The filesystem traversal benchmark needed better diagnostics to identify
performance bottlenecks in Eden and understand file access patterns. Without
detailed breakdowns, it was difficult to determine whether slowness was due
to small file overhead, large file I/O, or specific directory hotspots.

This adds a --detailed-read-stats flag that provides:
- File size distribution across 7 buckets (<1KB to >100MB)
- Small/medium/large file performance breakdown (10KB and 1MB thresholds)
- Per-category metrics showing time spent in open() vs read()
- Top 20 slowest directories with throughput analysis
- Directory depth distribution relative to traversal base paths

Key design decisions:
- Enum-indexed arrays for file categories ensure type safety
- Paths displayed relative to --dir arguments for cleaner output
- Lock-free atomics for thread-safe concurrent statistics updates
- Memory overhead: ~8 KB for histograms + unbounded HashMap for per-directory stats

Reviewed By: muirdm

Differential Revision: D84996648

fbshipit-source-id: ced2b9e151002719a5c3b20b1d104ccf335d788f

Author: Aaron Kushner

eden/fs/cli_rs/edenfs-commands/src/debug/bench/cmd.rs

@@ -117,6 +117,14 @@ pub enum BenchCmd {
             long_help = "Skip the file reading benchmark and only measure directory traversal performance. Useful for testing pure filesystem traversal speed without I/O overhead."
         )]
         skip_read: bool,
+
+        /// Show detailed read statistics including file size distribution and per-directory analysis
+        #[clap(
+            long,
+            help = "Show detailed read performance statistics",
+            long_help = "Enable detailed read statistics showing file size distribution, per-directory I/O performance breakdown, depth analysis, and file category overhead metrics. Only available when file reading is enabled (not compatible with --skip-read)."
+        )]
+        detailed_read_stats: bool,
     },
 }
 
@@ -196,7 +204,18 @@ impl crate::Subcommand for BenchCmd {
                 resource_usage,
                 json,
                 skip_read,
+                detailed_read_stats,
             } => {
+                // Validate flag compatibility
+                if *skip_read && *detailed_read_stats {
+                    return Err(anyhow::anyhow!(
+                        "--skip-read and --detailed-read-stats are mutually exclusive.\n\
+                        Detailed read statistics focus on file I/O performance metrics which are not \
+                        collected when skipping file reads. Use --detailed-read-stats without --skip-read \
+                        to see I/O performance analysis."
+                    ));
+                }
+
                 if !*json {
                     if dir.len() == 1 {
                         println!(
@@ -223,6 +242,7 @@ impl crate::Subcommand for BenchCmd {
                     *resource_usage,
                     *skip_read,
                     thrift_io.as_deref(),
+                    *detailed_read_stats,
                 )
                 .await?;