editoast: downgrade reqwest to mitigate timeout issues

Somewhere between 0.11.27 and 0.12.3, a regression related to timeout handling was introduced.
Other users seem to have similar issues: seanmonstar/reqwest#2283
Interestingly enough, this bug report mentions http/2, but the server we're querying only supports http/1.1.

Author: multun

editoast/Cargo.lock

@@ -94,7 +94,9 @@ redis = { version = "0.25.3", features = [
   "tokio-comp",
   "tokio-native-tls-comp",
 ] }
-reqwest = { version = "0.12.3", features = ["json"] }
+# 0.12.0 to 0.12.4 have weird timeout issues https://github.com/seanmonstar/reqwest/issues/2283
+# This bug was introduced between 0.12.0 and 0.12.3.
+reqwest = { version = "0.11.27", features = ["json"] }
 sentry = "0.32.3"
 sentry-actix = "0.32.3"
 serde.workspace = true