task: Add task contexts (#8675)

This PR supplements tasks with additional environment variables; ideally
we'll be able to write a task like:
`cargo test -p $ZED_CURRENT_PACKAGE -- $ZED_CURRENT_FUNCTION`
- [x] Flesh out multibuffer interactions
- [x] Add ZED_SYMBOL detection based on tree-sitter queries
- [ ] Add release note and demo
- [x] Figure out a solution for rerun dilemma - should `task: rerun`
reevaluate contexts for tasks?

This PR introduced the following variables:
- ZED_COLUMN - current line column
- ZED_ROW - current line row
and the following, which are available for buffers with associated
files:
- ZED_WORKTREE_ROOT - absolute path to the root of the current worktree.
- ZED_FILE - absolute path to the file
- ZED_SYMBOL - currently selected symbol; should match the last symbol
shown in a symbol breadcrumb (e.g. `mod tests > fn test_task_contexts`
should be equal to ZED_SYMBOL of `test_task_contexts`). Note that this
isn't necessarily a test function or a function at all.

Also, you can use them in `cwd` field of definitions (note though that
we're using https://docs.rs/subst/latest/subst/#features for that, so
don't expect a full shell functionality to work); the syntax should
match up with your typical Unix shell.


Release Notes:

- Added task contexts, which are additional environment variables set by
Zed for task execution; task content is dependent on the state of the
editor at the time the task is spawned.

---------

Co-authored-by: Anthony <anthonyeid7@protonmail.com>

crates/language/src/language_registry.rs

@@ -1,6 +1,6 @@
 use crate::{
-    CachedLspAdapter, Language, LanguageConfig, LanguageId, LanguageMatcher, LanguageServerName,
-    LspAdapter, LspAdapterDelegate, PARSER, PLAIN_TEXT,
+    CachedLspAdapter, Language, LanguageConfig, LanguageContextProvider, LanguageId,
+    LanguageMatcher, LanguageServerName, LspAdapter, LspAdapterDelegate, PARSER, PLAIN_TEXT,
 };
 use anyhow::{anyhow, Context as _, Result};
 use collections::{hash_map, HashMap};
@@ -78,6 +78,7 @@ struct AvailableLanguage {
     matcher: LanguageMatcher,
     load: Arc<dyn Fn() -> Result<(LanguageConfig, LanguageQueries)> + 'static + Send + Sync>,
     loaded: bool,
+    context_provider: Option<Arc<dyn LanguageContextProvider>>,
 }
 
 enum AvailableGrammar {
@@ -188,6 +189,7 @@ impl LanguageRegistry {
             config.name.clone(),
             config.grammar.clone(),
             config.matcher.clone(),
+            None,
             move || Ok((config.clone(), Default::default())),
         )
     }
@@ -237,6 +239,7 @@ impl LanguageRegistry {
         name: Arc<str>,
         grammar_name: Option<Arc<str>>,
         matcher: LanguageMatcher,
+        context_provider: Option<Arc<dyn LanguageContextProvider>>,
         load: impl Fn() -> Result<(LanguageConfig, LanguageQueries)> + 'static + Send + Sync,
     ) {
         let load = Arc::new(load);
@@ -257,6 +260,8 @@ impl LanguageRegistry {
             grammar: grammar_name,
             matcher,
             load,
+
+            context_provider,
             loaded: false,
         });
         state.version += 1;
@@ -422,6 +427,7 @@ impl LanguageRegistry {
                             .spawn(async move {
                                 let id = language.id;
                                 let name = language.name.clone();
+                                let provider = language.context_provider.clone();
                                 let language = async {
                                     let (config, queries) = (language.load)()?;
 
@@ -431,7 +437,9 @@ impl LanguageRegistry {
                                         None
                                     };
 
-                                    Language::new_with_id(id, config, grammar).with_queries(queries)
+                                    Language::new_with_id(id, config, grammar)
+                                        .with_context_provider(provider)
+                                        .with_queries(queries)
                                 }
                                 .await;