docs: Expand "wrt" to "with respect to".

This makes things more clear as this abbreviation isn't known
to everyone.

While the trailing period is common in French, it isn't in English
(one might use "w.r.t.").

Fixes #498.

src/pipeline/physics_hooks.rs

@@ -95,7 +95,8 @@ impl<'a> ContactModificationContext<'a> {
             CONTACT_CURRENTLY_FORBIDDEN => {
                 // Contacts are forbidden so we need to continue forbidding contacts
                 // until all the contacts are non-penetrating again. In that case, if
-                // the contacts are OK wrt. the contact normal, then we can mark them as allowed.
+                // the contacts are OK with respect to the contact normal, then we can
+                // mark them as allowed.
                 if contact_is_ok && self.solver_contacts.iter().all(|c| c.dist > 0.0) {
                     *self.user_data = CONTACT_CURRENTLY_ALLOWED;
                 } else {

src/pipeline/physics_pipeline.rs

@@ -630,8 +630,8 @@ impl PhysicsPipeline {
         }
 
         // Finally, make sure we update the world mass-properties of the rigid-bodies
-        // that moved. Otherwise, users may end up applying forces wrt. an outdated
-        // center of mass.
+        // that moved. Otherwise, users may end up applying forces with respect to an
+        // outdated center of mass.
         // TODO: avoid updating the world mass properties twice (here, and
         //       at the beginning of the next timestep) for bodies that were
         //       not modified by the user in the mean time.

src/pipeline/query_pipeline.rs

@@ -706,7 +706,7 @@ impl QueryPipeline {
     /// * `stop_at_penetration` - If the casted shape starts in a penetration state with any
     ///    collider, two results are possible. If `stop_at_penetration` is `true` then, the
     ///    result will have a `toi` equal to `start_time`. If `stop_at_penetration` is `false`
-    ///    then the nonlinear shape-casting will see if further motion wrt. the penetration normal
+    ///    then the nonlinear shape-casting will see if further motion with respect to the penetration normal
     ///    would result in tunnelling. If it does not (i.e. we have a separating velocity along
     ///    that normal) then the nonlinear shape-casting will attempt to find another impact,
     ///    at a time `> start_time` that could result in tunnelling.