[benchmark] Make paranoid mode configurable for replay (#15900)

aptos-move/replay-benchmark/README.md

@@ -211,9 +211,12 @@ of transactions. Typically, you want to have the concurrency level to match the
 multiple concurrency levels are provided, the benchmark reports the measurements for each level.
 This way it is possible to see how concurrency affects the runtime. 
 
-Finally, in order to differentiate between cold and warm starts, there is an option to skip the
-measurement for the first few blocks. By specifying `--num-block-to-skip N`, the tool will ignore
-measurements for the first `N` blocks (the blocks will still be executed as a "warm-up").
+In order to differentiate between cold and warm starts, there is an option to skip the measurement
+for the first few blocks. By specifying `--num-block-to-skip N`, the tool will ignore measurements
+for the first `N` blocks (the blocks will still be executed as a "warm-up").
+
+Execution can also be configured. By using `--disable-paranoid-mode`, the Move VM will not use
+runtime type checks, possible making execution faster.
 
 #### Example
 

aptos-move/replay-benchmark/src/commands/benchmark.rs

@@ -9,6 +9,7 @@ use crate::{
 };
 use anyhow::{anyhow, bail};
 use aptos_logger::Level;
+use aptos_vm_environment::prod_configs::set_paranoid_type_checks;
 use clap::Parser;
 use std::path::PathBuf;
 use tokio::fs;
@@ -61,6 +62,13 @@ pub struct BenchmarkCommand {
                 the overall time to execute all blocks"
     )]
     measure_overall_time: bool,
+
+    #[clap(
+        long,
+        default_value_t = false,
+        help = "If false, Move VM runs in paranoid mode, if true, paranoid mode is not used"
+    )]
+    disable_paranoid_mode: bool,
 }
 
 impl BenchmarkCommand {
@@ -109,6 +117,7 @@ impl BenchmarkCommand {
             })
             .collect::<Vec<_>>();
 
+        set_paranoid_type_checks(!self.disable_paranoid_mode);
         BenchmarkRunner::new(
             self.concurrency_levels,
             self.num_repeats,