Create benchmark stage in the pipeline

- Unfortunatelly the pennylane [benchmark](https://github.com/PennyLaneAI/pennylane/tree/master/benchmark) folder is not available in the pip install version of pennylane. Therefore [pennylane](https://github.com/PennyLaneAI/pennylane) is added as submodule to the project.
- Created benchmark stage, the result of the benchmark can be downloaded from the pipeline.
- More info in the [README](README.md#benchmarking)

.gitlab-ci.yml

@@ -1,5 +1,11 @@
 image: python:3.6
 
+variables:
+  GIT_SUBMODULE_STRATEGY: normal #to use submodule
+
+before_script:
+  - pip install poetry
+
 stages:
   - test
   - build
@@ -8,16 +14,29 @@ unittest:
   stage: test
   script:
     - pip install tox
-    - pip install poetry
     - tox
 
+benchmark:
+    stage: test
+    script:
+      - apt update
+      - apt-get install libeigen3-dev
+      - poetry install
+      - cd pennylane/benchmark/
+      - poetry run python benchmark.py -d default.qubit time bm_entangling_layers > ./output.txt
+    artifacts:
+      paths:
+        - pennylane/benchmark/output.txt
+      expire_in: 15 days
+
+
 buildpackage:
   stage: build
   only: 
     - merge_requests
   script:
-    - pip install poetry
     - poetry config repositories.gitlab https://gitlab.inf.elte.hu/api/v4/projects/73/packages/pypi
     - poetry config pypi-token.gitlab 1z5AF-A6zsitDWpwGP91
     - poetry build
     - poetry publish -r gitlab
+

.gitmodules

@@ -0,0 +1,3 @@
+[submodule "pennylane"]
+	path = pennylane
+	url = https://github.com/XanaduAI/pennylane.git

README.md

@@ -19,7 +19,7 @@ on a machine running Fedora/CentOS (the rpm package is named `eigen3-devel`).
 sudo apt install python3-venv
 ```
 
-Additionally, this project uses `tox` to manage virtualenvs, install it with 
+Additionally, this project uses `tox` to manage virtualenvs, install it with:
 ``` 
 pip install tox
 ```
@@ -100,3 +100,19 @@ https://python-poetry.org/docs/libraries/#publishing-to-a-private-repository
 ```
 poetry publish -r gitlab
 ```
+## Benchmarking
+We can run benckmarks with various pennylane devices, for more details see:
+https://github.com/XanaduAI/pennylane/blob/master/benchmark/README.rst
+For instance:
+```
+cd pennylane/benchmark/
+poetry run python benchmark.py -d default.qubit time bm_entangling_layers
+```
+### Notes:
+Unfortunatelly running `benchmark.py` with `bm_mutable_rotations` or
+`bm_mutable_complicated_params` will not work. There is a bug in both of these
+in their `__init__()` function: they are expected to accept a named parameter
+`qnode_type` but they don't. If you add the parameter by hand it will work
+fine.
+
+**The benchmarks, as of now, only work for qubit operations!**