Clean up and improve release profile flags

Cargo.lock

@@ -76,18 +76,18 @@ dependencies = [
 
 [[package]]
 name = "clap"
-version = "4.5.32"
+version = "4.5.34"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "6088f3ae8c3608d19260cd7445411865a485688711b78b5be70d78cd96136f83"
+checksum = "e958897981290da2a852763fe9cdb89cd36977a5d729023127095fa94d95e2ff"
 dependencies = [
  "clap_builder",
 ]
 
 [[package]]
 name = "clap_builder"
-version = "4.5.32"
+version = "4.5.34"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "22a7ef7f676155edfb82daa97f99441f3ebf4a58d5e32f295a56259f1b6facc8"
+checksum = "83b0f35019843db2160b5bb19ae09b4e6411ac33fc6a712003c33e03090e2489"
 dependencies = [
  "anstyle",
  "clap_lex",
@@ -262,9 +262,9 @@ dependencies = [
 
 [[package]]
 name = "once_cell"
-version = "1.21.1"
+version = "1.21.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d75b0bedcc4fe52caa0e03d9f1151a323e4aa5e2d78ba3580400cd3c9e2bc4bc"
+checksum = "42f5e15c9953c5e4ccceeb2e7382a716482c34515315f7b03532b8b4e8393d2d"
 
 [[package]]
 name = "oorandom"

Cargo.toml

@@ -7,16 +7,20 @@ edition = "2024"
 debug-layout = []
 debug-latency = []
 
+# `opt-level = "s"` may be useful in the future as it significantly reduces binary size.
+# The primary problem are the ucd and utf8 modules which take a 20% performance hit.
 [profile.release]
-codegen-units = 1
-debug = "full"
-lto = true
-panic = "abort"
-debug-assertions = true # Temporary while I test this
+codegen-units = 1           # reduces binary size by ~2%
+debug = "full"              # No one needs an undebuggable release binary
+debug-assertions = true     # TODO: Temporary while I test this
+lto = true                  # reduces binary size by ~14%
+panic = "abort"             # reduces binary size by ~50% in combination with -Zbuild-std-features=panic_immediate_abort
+split-debuginfo = "packed"  # generates a seperate *.dwp/*.dSYM so the binary can get stripped
+strip = "symbols"           # See split-debuginfo - allows us to drop the size by ~65%
 
 [profile.bench]
-codegen-units = 16
-lto = "thin"
+codegen-units = 16          # Make compiling criterion faster (16 is the default, but profile.release sets it to 1)
+lto = "thin"                # Similarly, speed up linking by a ton
 
 [dependencies]