After hours and hours (and hours) of life sucking debugging I found
out the offending code (in read_key_sequence):

    if ((FIXNUMP (key) && XFIXNUM (key) == -2) /* wrong_kboard_jmpbuf */
/* When switching to a new tty (with a new keyboard),
   read_char returns the new buffer, rather than -2
   (Bug#5095).  This is because `terminal-init-xterm'
   calls read-char, which eats the wrong_kboard_jmpbuf
   return.  Any better way to fix this? -- cyd  */
|| (interrupted_kboard != current_kboard))
  {
   ....
   mock_input = 0;
   goto replay_entire_sequence;
  }

The problem with the fix for #5095 is that it's dropping the current
event, which is a valid char event and not -2. Thus, the first esc
char sent by the terminal after turning on focus tracking is lost and
we get `[`, `I` events separately inserted into the buffer afterwards.

Attached is a patch that replays key as mock input when it's not -2.

Can I ask for some of my hidpi fringe or colorized comint/python
history patches to be merged in exchange ;) ?

Best regards
--
Carlos