Commit d1e0e3f introduced a `remove_nonprint` method that breaks hamlib on
all TM-D710/TM-V71A devices by erroneously removing the command termination
character (`\r`).

This commit adopts a solution proposed by @GeoBaltz that only runs
`remove_nonprint` if the command termination character itself is printable.

Resolves: #1767 #1698
(cherry picked from commit 85c9e15)

set_chan() was correctly creating and sending the command, and returning RIG_OK. However, radio was actually ignoring it because command wasn't terminated with a CR. This is now corrected.

(cherry picked from commit aa39d6a)

strip_quotes() orphaned 1 or 2 strings per call.

(cherry picked from commit 0740af6)

(cherry picked from commit 65d922c)

Response to rigctld command was being corrupted by other threads
Fixes issue #1748

(cherry picked from commit 8d0e67f)

The R75 for some reason rejects the powerstat query and returns an error. Commented out  .get_powerstat to correct that. Applications should initially assume it's on, then internally track power status, since you can still turn it off.

(cherry picked from commit dc12b01)

(cherry picked from commit ec4590d)

rot_sprintf_status() was printing items multiple times.
Cause of problem noticed by cppcheck.

(cherry picked from commit 0b75b96)

Added the option to send CW using CWX commands on Flex's.

(cherry picked from commit b365d98)

(cherry picked from commit 3b65778)

Needed to replace spaces with 0x7f characters.

(cherry picked from commit 0c94ba4)