Telephone Keypad String Encoding

Encode text using a traditional telephone keypad. Letters on keys 2 through 9 require one through four presses according to their position on the key.
Ignore spaces and treat uppercase and lowercase letters identically.
If two consecutive encoded letters use the same key, insert # between their digit sequences.
Return the complete encoded string.

Function
encodeKeypad(text: String) → String

Examples
Example 1
text = "AB"
return = "2#22"
A and B share key 2, so a separator is inserted.
Example 2
text = "HEY EPIC"
return = "4433999337444222"
The space is ignored and no adjacent encoded letters share a keypad key.

Constraints
1 <= text.length <= 100000
text contains English letters and spaces.
text contains at least one letter.

The trick: map each letter to its key digit and its press count, then build the output by walking the string once. Lowercase everything first, skip spaces, and track the last key you emitted. If the current letter's key equals the last key, append # before the digits. Then set last key to the current one. The common pitfall is resetting the previous key when you hit a space. Don't. The spec says spaces are ignored, so 'A B' has A and B adjacent and needs the separator. Example 1 shows this: AB gives 2#22. Another pitfall is string concatenation in a loop. With up to 100000 characters, use a list or StringBuilder and join at the end. Remember the keypad: 7 (PQRS) and 9 (WXYZ) have four letters, the rest have three. If you blank on the table layout during the live OA, StealthCoder can cover you, but the logic is just one pass and O(n).

Drill it cold or hedge it with StealthCoder. Either way, don't walk into the OA hoping you remember the trick.