Add client and server ts information to websocket messages to surface websocket latency issues (#40827)

This PR adds the `clientTs` to the initial websocket connect message, which the sync worker uses to calculate the `client_clock_skew`, which is the difference between the client clock and server clock, plus whatever latency between client sending and server receiving the message.

We send the latest possible `serverTs` before the server `Transition` message gets sent down to the client. On the client, we can then understand how long it took for the `Transition` message to download by comparing the current client time with the `serverTs` and subtracting the `clientClockSkew`.

GitOrigin-RevId: 9b8d2693a40196a78ee08aa3e2f49c695d0172d0

Author: emmaling27

src/browser/sync/client.ts

@@ -396,6 +396,7 @@ export class BaseConvexClient {
             type: "Connect",
             sessionId: this._sessionId,
             maxObservedTimestamp: this.maxObservedTimestamp,
+            clientTs: Date.now(),
           });
 
           // Throw out our remote query, reissue queries

src/browser/sync/protocol.ts

@@ -121,6 +121,7 @@ type Connect = {
   connectionCount: number;
   lastCloseReason: string | null;
   maxObservedTimestamp?: TS;
+  clientTs: number;
 };
 
 export type AddQuery = {
@@ -254,6 +255,8 @@ export type Transition = {
   startVersion: StateVersion;
   endVersion: StateVersion;
   modifications: StateModification[];
+  clientClockSkew?: number;
+  serverTs?: number;
 };
 
 type MutationSuccess = {