Fix multiple Vale process spawning issue (#79)

This fixes the regression of the issue originally resolved in PR #18,
where multiple Vale processes would spawn and consume excessive CPU.

The LSP-based rewrite (PR #41) removed the old safeguards. This commit
adds proper client lifecycle management:

- Check for existing client in activate() and stop it before creating new one
- Improve deactivate() with async/await and error handling
- Add logging for better debugging of client lifecycle

This ensures only one vale-ls process runs at a time, preventing the
CPU spike issues users were experiencing.

Fixes: Issue with multiple Vale processes spawning
Related: PR #18

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-authored-by: Claude <[email protected]>

Author: theletterf

src/lsp.ts

@@ -92,6 +92,16 @@ interface valeArgs {
 }
 
 export async function activate(context: ExtensionContext) {
+  // Prevent multiple activations - stop existing client if present
+  if (client) {
+    console.log("Vale language client already active, stopping existing client");
+    try {
+      await client.stop();
+    } catch (error) {
+      console.error("Error stopping existing client:", error);
+    }
+  }
+
   let filePath = path.join(context.extensionPath, "vale-ls");
 
   // Handle Windows
@@ -214,9 +224,15 @@ export async function activate(context: ExtensionContext) {
   }
 }
 
-export function deactivate(): Thenable<void> | undefined {
+export async function deactivate(): Promise<void> {
   if (!client) {
-    return undefined;
+    return;
+  }
+
+  try {
+    await client.stop();
+    console.log("Vale language server stopped");
+  } catch (error) {
+    console.error("Error stopping Vale language server:", error);
   }
-  return client.stop();
 }