Add onLongPress to TextInput - #59148

Summary:

TextInput already handles its touches through Pressability and exposes onPress, onPressIn and onPressOut, but not onLongPress, so a component cannot tell a held press on a field from a tap. #39701 added onPress on the same grounds.

This passes onLongPress to the Pressability config TextInput already builds. A touch held past the long press delay calls it, and the release that follows calls no onPress, as on Pressable. That release is where TextInput focuses itself, so the input is focused when onLongPress fires instead; without that, a held field with contextMenuHidden on Android ends focused by the platform with no keyboard. Without onLongPress, a held press is still a press.

I need it for a one-time-code field that sets contextMenuHidden on Android and offers its own Paste on a long press.

Changelog:

[GENERAL] [ADDED] - Add onLongPress to TextInput

Test Plan:

Four Fantom cases in TextInput-itest.js under onLongPress: a touch held past the delay calls it once; it focuses the input when it fires, and the release that follows calls no onPress; a press released before the delay still calls onPress and focuses; and without onLongPress a held press still calls onPress and focuses. On main the first two fail (onLongPress is called 0 times; no focus command is sent when the delay passes); with this change the file passes, 38 of 38. yarn flow-check, yarn lint, yarn test-typescript-legacy, yarn build-types (the ReactNativeApi.d.ts snapshot gains onLongPress on TextInputBaseProps), yarn test-generated-typescript and yarn test-ci pass. Mounting and updating 100 TextInputs in a Fantom benchmark, with and without onLongPress, costs the same on main as with this change, within the spread between runs.

On an iPhone 12 Pro Max (iOS 27.0.1) with Xcode 27.1, where CI pins 16.4, RNTester's Debug and Release builds pass. So do 159 of RNTesterUnitTests' 179 tests, with 10 skipped. The 10 that fail also fail on main: four RCTBundleURLProviderTests that read the device's packager setup, RCTConvert_NSColorTests testGenerateFallbacks on iOS 27's colors, and the five RCTBlobManagerTests, whose setUp still sets the bridge key that #59250 stopped synthesizing. Visual regression (ios) reports switch/custom-colors/toggled changed by 11804 px, the same difference main's own runs report on da42db9c90, f9340f07dc and 11b10c0913.

In RNTester's Playground:

function Playground() {
  const [events, setEvents] = React.useState([]);
  const record = name => setEvents(previous => [...previous, name]);
  return (
    <View style={styles.container}>
      <TextInput
        placeholder="Hold me"
        onLongPress={() => record('onLongPress')}
        onPress={() => record('onPress')}
      />
      <TextInput
        contextMenuHidden={true}
        placeholder="Hold me, menu hidden"
        onLongPress={() => record('onLongPress (menu hidden)')}
        onPress={() => record('onPress (menu hidden)')}
      />
      <RNTesterText>{events.join(', ')}</RNTesterText>
    </View>
  );
}

Holding each field for 1.5 s and lifting, in RNTester debug builds on an Android 16 (API 36) emulator: on main nothing is logged while the finger is down and onPress is logged on release; with this change onLongPress is logged while it is down and nothing on release. Both fields end focused with the keyboard up on both builds, and a tap logs onPress on both. Holding a contextMenuHidden field on its text leaves the keyboard down on both builds, because Android's own long press focuses the field first; #59130 stops that long press.