subscriber: impl Clone for EnvFilter

This is useful when using `EnvFilter` for multiple identical per-layer
filters, as well as with clap and similar libraries that have `Clone`
bounds.

We generally expect users to be cloning an `EnvFilter` before attaching it
to a subscriber, rather than cloning `EnvFilters` that are already
attached. Because of this, we reset all the accumulated dynamic state
when cloning. This means that some spans and callsites might be missed
when an already-attached `EnvFilter` is cloned, but the presence of the
dynamic state mean that detaching and attaching `EnvFilter`s to existing
subscribers (e.g. with `reload`) already doesn't work very well. This
isn't a new class of problem.

There was a previous implementation of this in tokio-rs#2398, that shared the
dynamic state between all cloned filters behind an `Arc`. I chose
not do go for that approach because it causes inconsistencies if the
cloned filters are attached to different subscribers.

Fixes: tokio-rs#2360

Author: olivia-fl

tracing-subscriber/src/filter/env/mod.rs

@@ -204,6 +204,24 @@ pub struct EnvFilter {
     regex: bool,
 }
 
+/// Creates an [`EnvFilter`] with the same directives as `self`.
+///
+/// This does *not* clone any of the dynamic state that [`EnvFilter`] acquires while attached to a
+/// subscriber.
+impl Clone for EnvFilter {
+    fn clone(&self) -> EnvFilter {
+        EnvFilter {
+            statics: self.statics.clone(),
+            dynamics: self.dynamics.clone(),
+            has_dynamics: self.has_dynamics,
+            by_id: RwLock::default(),
+            by_cs: RwLock::default(),
+            scope: ThreadLocal::new(),
+            regex: self.regex,
+        }
+    }
+}
+
 type FieldMap<T> = HashMap<Field, T>;
 
 /// Indicates that an error occurred while parsing a `EnvFilter` from an