refactor: move text-to-speech out of chromium_src (#15024)

* chore: add tts patch and buildflag, makes tts work again

* chore: add tts patch and buildflag, makes tts work again

* fix: make things compile

* build: add relevant tts files for linux

* fix: update patch and patch description, should now compile on mac

* build: move chrome specific sources under chromium_src:chrome target

* build: enable_extensions again

We are depending on them, check `//electron/chromium_src:chrome` target
for more info.

* fix: update tts.patch to receive notifications about browser context destruction

* fix: extend browser process from chrome layer

The global state g_browser_process is shared between //chrome
and //electron.

* spec: add basic speech synthesis test

* spec: skip speech tests on ci

* build: fix compilation on windows

spec/chromium-spec.js

@@ -1,4 +1,6 @@
 const assert = require('assert')
+const chai = require('chai')
+const dirtyChai = require('dirty-chai')
 const fs = require('fs')
 const http = require('http')
 const path = require('path')
@@ -12,6 +14,9 @@ const { app, BrowserWindow, ipcMain, protocol, session, webContents } = remote
 const isCI = remote.getGlobal('isCi')
 const features = process.atomBinding('features')
 
+const { expect } = chai
+chai.use(dirtyChai)
+
 /* Most of the APIs here don't use standard callbacks */
 /* eslint-disable standard/no-callback-literal */
 
@@ -1269,4 +1274,41 @@ describe('chromium feature', () => {
       })
     })
   })
+
+  describe('SpeechSynthesis', () => {
+    before(function () {
+      if (isCI || !features.isTtsEnabled()) {
+        this.skip()
+      }
+    })
+
+    it('should emit lifecycle events', async () => {
+      const sentence = `long sentence which will take at least a few seconds to
+          utter so that it's possible to pause and resume before the end`
+      const utter = new SpeechSynthesisUtterance(sentence)
+      // Create a dummy utterence so that speech synthesis state
+      // is initialized for later calls.
+      speechSynthesis.speak(new SpeechSynthesisUtterance())
+      speechSynthesis.cancel()
+      speechSynthesis.speak(utter)
+      // paused state after speak()
+      expect(speechSynthesis.paused).to.be.false()
+      await new Promise((resolve) => { utter.onstart = resolve })
+      // paused state after start event
+      expect(speechSynthesis.paused).to.be.false()
+
+      speechSynthesis.pause()
+      // paused state changes async, right before the pause event
+      expect(speechSynthesis.paused).to.be.false()
+      await new Promise((resolve) => { utter.onpause = resolve })
+      expect(speechSynthesis.paused).to.be.true()
+
+      speechSynthesis.resume()
+      await new Promise((resolve) => { utter.onresume = resolve })
+      // paused state after resume event
+      expect(speechSynthesis.paused).to.be.false()
+
+      await new Promise((resolve) => { utter.onend = resolve })
+    })
+  })
 })