I just got my new LBB and I'm also having the same issues on raspberry pi 3 Jessie. My tool versions and symptoms are the same as the TP.
The SDK firmware (201611070000) is newer than what's on the bean (201609290000) and flashing hangs. I can load and run sketchs from the android loader (which I did first before trying the linux environment) and will note that on my first run of the android app, it did a LBB firmware update. It didn't say (or I didn't notice) what version was being replaced.
Can we get some folks from PunchThrough to chime in?
Edit: I have an additional symptom that may be helpful in isolating the issue. At those times when the linux CLI is hung in a command (loading sketch, firmware update, log_serial, etc.) the android loader is unable to contact the LBB. Once the CLI is terminated, the android loader can once again make contact. So, I presume the CLI is getting far enough to establish and hold a connection.
Edit2: Yet another symptom. When I load a sketch from android that does serial output and then use the linux CLI in an attempt to read it, after I kill the CLI with ^C^C, when I do a CLI scan, I often get the output:
pi@pi-32:/usr/bin $ bean scan
Scanning for LightBlue devices for 30 seconds...
New Device discovered or updated!
noble warning: unknown handle 64 disconnected!
No doubt the unknown handle is associated with the previous hung CLI command. What I'm not sure of is was that warning message generated from info returned by the LBB or some other process on the raspberry pi. Is there a server process on the host system involved with LLB communication?