style(rust): drop the comments that narrated this code's history

A comment describes the code as it is. The history belongs in the commit message,
where it is attached to the diff that makes it true; in the source it is stale
for everyone who never saw the before.

The comment above the `timeout` parse goes entirely: the field's doc comment
already states the default. The zero rate-limit check states its constraint
instead, and the three test comments say what they pin.

Reported on the pull request as "No need to state the history of this code."

Author: wkirschenmann

packages/rust/armonik-transport/src/config.rs

@@ -287,9 +287,6 @@ impl ClientConfig {
             )
         };
 
-        // `None`, not `Some(60s)`: nothing applied this until now, so the 60s was never observable, and
-        // keeping it while starting to apply it would deadline every existing caller's requests at one
-        // minute. `connect_timeout` above is the opposite case, its 60s having always been applied.
         let timeout = if timeout.is_empty() {
             None
         } else {
@@ -677,8 +674,8 @@ mod tests {
 
     #[test]
     fn a_zero_rate_limit_is_rejected_rather_than_left_to_panic() {
-        // `tower`'s `Rate::new` asserts both halves are above zero, and this is the change that starts
-        // handing them to it: without this, `0/1s` panics inside `connect` instead of being reported.
+        // `tower`'s `Rate::new` asserts both halves are above zero, so a zero has to be refused here
+        // rather than reaching it: a panic inside `connect` tells the caller nothing.
         for value in ["0/1s", "1/0s", "0/0s"] {
             let error = ClientConfig::from_config_args(ClientConfigArgs {
                 rate_limit: String::from(value),

packages/rust/armonik-transport/tests/timeout.rs

@@ -1,7 +1,7 @@
-//! `GrpcClient__Timeout` and `GrpcClient__RateLimit`, now that they reach the channel.
+//! `GrpcClient__Timeout` and `GrpcClient__RateLimit` reaching the channel.
 //!
-//! A test that only asserted the parsing would have passed before this change too, since both options
-//! were already parsed and then dropped. So these go through a real connection to a real server.
+//! Through a real connection to a real server, measuring what the caller gets: asserting on the parsing
+//! alone would say nothing about whether either option reaches the channel.
 
 mod common;
 
@@ -11,8 +11,7 @@ use common::{call, config, serve, SlowService};
 
 #[tokio::test]
 async fn a_request_timeout_ends_a_call_the_server_is_too_slow_to_answer() {
-    // The server takes ten seconds; the caller allows 300ms. Before this change the option was parsed
-    // and ignored, so the call waited for the full ten.
+    // A server that takes ten seconds against a caller that allows 300ms.
     let endpoint = serve(SlowService::new(Duration::from_secs(10))).await;
 
     let channel = armonik_transport::connect(config(&endpoint, |args| {
@@ -76,9 +75,7 @@ async fn a_rate_limit_is_accepted_and_still_lets_calls_through() {
 
 #[test]
 fn an_empty_timeout_means_no_timeout_rather_than_a_minute() {
-    // The field is documented as defaulting to no timeout, and nothing applied it until now, so the
-    // `Some(60s)` it used to parse to was never observable. This is the assertion that keeps a
-    // one-minute deadline from appearing on every request of every caller.
+    // What keeps a one-minute deadline off every request of every caller who set nothing.
     let config = config("http://localhost:5001", |_| {});
 
     assert_eq!(config.timeout, None);