Reduce CircleCI threads, add libglfw3 for AF in CUDA Docker build

Summary:
CircleCI hosts are OOM/build processes are being killed because we're using too many threads; reduce to 1.

We need `libglfw3` as of ArrayFire v3.6.4.

Reviewed By: tlikhomanenko

Differential Revision: D15545415

fbshipit-source-id: defe8acc8ec71d76c1605de70a3f4b1fb9dbdcd1

.circleci/config.yml

@@ -16,7 +16,7 @@ jobs:
           command: |
             mkdir -p build && cd build
             cmake .. -DCMAKE_BUILD_TYPE=Release -DFLASHLIGHT_BACKEND=CUDA
-            make -j8 && make install
+            make && make install
   build-cpu:
     docker:
       - image: flml/flashlight:cpu-base-latest
@@ -26,4 +26,4 @@ jobs:
           command: |
             export MKLROOT=/opt/intel/mkl && mkdir -p build && cd build
             cmake .. -DCMAKE_BUILD_TYPE=Release -DFLASHLIGHT_BACKEND=CPU
-            make -j8 && make install
+            make && make install

Dockerfile-CUDA-Base

@@ -32,7 +32,8 @@ RUN APT_INSTALL="apt-get install -y --no-install-recommends" && \
         # OpenMPI
         libopenmpi-dev libomp-dev \
         # nccl: for flashlight
-        libnccl2 libnccl-dev && \
+        libnccl2 libnccl-dev \
+        libglfw3-dev && \
 # ==================================================================
 # arrayfire https://github.com/arrayfire/arrayfire/wiki/
 # ------------------------------------------------------------------