This is a curious thing in deed: 

open one user A (with BT support) in plasma
open another user B (with BT support in plasma.

User A is bound to tty2
User B is bound to tty3 (or if you do some error it shifts to tty7 as the new
behavior is like this if you use alt+cntrl+Fn to shift to another tty (and
forget that it is not yet open). 

Long story short, go to the second user session. It will connect problem free
to the BT phone or whatever you wish to do a file transfer. Start the sending
of the file on the e.g. Android phone. 
It will fail silently. 

Now shift with alt+cntr+Fn to the first(!) opened user session. Now the
transfer if it is still active on the phone will suddenly start, transferring
the file to the wrong user session. 
I also tried the following to avoid this: switch off BT in the first user
(A)session, switch to the second user session (B), switch there BT on, try to
transfer the file, fails silently, then go back to the first user session. BT
will be active as expected and here it will receive immediately the files. 

Hence this is a focus stealing issue. And when you look at dmesg it complains
of anything and lists only:
[17822.836978] [ T1445] input: Pan Galactic Gargle Blaster (AVRCP) as
/devices/virtual/input/input31
[17862.458018] [ T1445] input: Pan Galactic Gargle Blaster (AVRCP) as
/devices/virtual/input/input32
[17910.817347] [ T1445] input: Pan Galactic Gargle Blaster (AVRCP) as
/devices/virtual/input/input33
[18020.812264] [ T1445] input: Pan Galactic Gargle Blaster (AVRCP) as
/devices/virtual/input/input34

where Pan Galactic Gargle Blaster is the name of the phone (Poco X3 Pro).
In journalctl you find the following (very few) lines:

giu 03 14:21:00 silversurfer bluetoothd[1445]:
src/profile.c:ext_io_disconnected() Unable to get io data for Object Push:
getpeername: Transport endpoint is not connected (107)
giu 03 14:21:00 silversurfer obexd[2731]: disconnected: Transport got
disconnected
giu 03 14:21:00 silversurfer obexd[2731]: DISCONNECT(0x1), Success(0x20)
giu 03 14:21:00 silversurfer obexd[2731]: DISCONNECT(0x1), <unknown>(0xff)
giu 03 14:20:55 silversurfer bluetoothd[1445]:
/org/bluez/hci0/dev_8C_D9_D6_3B_1E_4F/fd3: fd(42) ready
giu 03 14:20:55 silversurfer systemd-logind[1504]: Watching system buttons on
/dev/input/event26 (Pan Galactic Gargle Blaster (AVRCP))


before there is only one line that speaks about a problem:
giu 03 14:19:33 silversurfer wireplumber[6733]:
org.bluez.GattManager1.RegisterApplication() failed:
GDBus.Error:org.bluez.Error.AlreadyExists: Already Exists

(end of excerpt from journal). 
You find this combination of errors for all aborted transfers. 


What the software does: getting an incoming file from a BT sender, the active
session should be the receiver, and NOT the first tty user session that was
opened that day. 

What the software does: aborts transfers to the user session if it is not the
first in chronological line of log in (first active tty for GUI). 

Software involved Bluez, bluetoothd and obex