fix(lsp): followup fixes for semantic tokens support (#21357)

1. The algorithm for applying edits was slightly incorrect. It needs to
   preserve the original token list as the edits are applied instead of
   mutating it as it iterates. From the spec:

   Semantic token edits behave conceptually like text edits on
   documents: if an edit description consists of n edits all n edits are
   based on the same state Sm of the number array. They will move the
   number array from state Sm to Sm+1.

2. Schedule the semantic token engine start() call in the
   client._on_attach() function so that users who schedule_wrap() their
   config.on_attach() functions (like nvim-lspconfig does) can still
   disable semantic tokens by deleting the semanticTokensProvider from
   their server capabilities.

runtime/lua/vim/lsp/semantic_tokens.lua

@@ -313,18 +313,15 @@ function STHighlighter:process_response(response, client, version)
       return a.start < b.start
     end)
 
-    ---@private
-    local function _splice(list, start, remove_count, data)
-      local ret = vim.list_slice(list, 1, start)
-      vim.list_extend(ret, data)
-      vim.list_extend(ret, list, start + remove_count + 1)
-      return ret
-    end
-
-    tokens = state.current_result.tokens
+    tokens = {}
+    local old_tokens = state.current_result.tokens
+    local idx = 1
     for _, token_edit in ipairs(token_edits) do
-      tokens = _splice(tokens, token_edit.start, token_edit.deleteCount, token_edit.data)
+      vim.list_extend(tokens, old_tokens, idx, token_edit.start)
+      vim.list_extend(tokens, token_edit.data)
+      idx = token_edit.start + token_edit.deleteCount + 1
     end
+    vim.list_extend(tokens, old_tokens, idx)
   else
     tokens = response.data
   end