Set machine-type of the PAPI worker (#567)

* Set machine-type for the runner worker

All this worker does is to launch the Dataflow pipeline, monitor the result, and finlay load AVRO files to the destination BQ tables.
Using a smaller machine would help our customers to save on extra cost of running this worker.

* Rollback #471
and use gcloud instead of pipelines tools. This will remove an extra dependency and also make our docker image smaller.
Here we also update our base image to cloud-sdk:286.0.0-slim

* Keep using piplines tools

Decided to not move back to gcloud command in this PR.

docker/Dockerfile

@@ -19,7 +19,7 @@ FROM golang:1.11
 
 RUN go get -ldflags '-extldflags "-fno-PIC -static"' -buildmode pie -tags 'osusergo netgo static_build' github.com/googlegenomics/pipelines-tools/pipelines
 
-FROM google/cloud-sdk:276.0.0-slim
+FROM google/cloud-sdk:286.0.0-slim
 
 COPY --from=0 /go/bin/pipelines /usr/bin
 
@@ -44,7 +44,8 @@ RUN apt-get update && apt-get install -y \
     libssl-dev \
     make \
     perl \
-    zlib1g-dev
+    zlib1g-dev \
+    python-pip
 
 # Install dependencies.
 RUN python -m pip install --upgrade pip && \

docker/pipelines_runner.sh

@@ -85,6 +85,7 @@ function main {
     --scopes "https://www.googleapis.com/auth/cloud-platform" \
     --regions "${region}" \
     --image "${vt_docker_image}" \
+    --machine-type "g1-small" \
     --pvm-attempts 0 \
     --attempts 1 \
     --disk-size 10