mulle: Update RTT config to match other Kinetis boards

frdm-k22f: Correct copy pasta in LEDx macros

at86rf2xx: Avoid deadlock on manual retransmissions with NETOPT_PRELOADING

tests: trickle: extend trickle test to print SUCCESS/FAILURE

kinetis: Generic ISR vector table for all Kinetis CPUs

The ISR names have been changed to match the name of the IRQ number they
are servicing.

According to NXP material, there are no 100 MHz K22F parts in 144 pin
packages, which is the reference manual that this header is supposed to
correspond to.
The header was originally included from Keil uVision Kinetis support
packages. It is possible that this header was only used with
engineering samples of the K22F.

tests/ssp: Fix on macOS while compiling w/ clang

On macOS using Apple LLVM version 9.0.0 this test would
not compile due to clang detecting the buffer overflow.
Since this test is all about having a buffer overflow, I
tricked clang into not detecting this anymore.

This loosely relates to #6473.

make: fix info-buildsizes-diff

tests/sx127x: remove unused periph/rtc.h include

boards: add support for bluepill board

dist: compile_test: cleanup after buildtest

[backport] Release notes of 2017.10

drivers/sx127x: Fix misc variable types and lengths

trickle: correctly reset t

tests/driver_servo/main.c: fix var len

pm_layered: make sure no conflict for pm_set_lowest exists

sys: libc: only add include path on non-native boards

Removed a comma from the Boards API reference.

We are at a point were building all of RIOT takes massive amounts of
storage. Since my machine runs full after only compiling ~20
applications (I have ~170 GB free storage if the RIOT repo is clean).

Thus the easiest solution for the compile tests is to just clean-up the
mess they made ;-).

tests/driver_isl29125/main.c: fix counter var type

tests/driver_adcxx1c: fix warning unused  argument

drivers/encx24j600/encx24j600.c : fix counter var length