Reduce compilation time of the QC project

[knopers8]

Starting off from #180 as a point of reference, to avoid too much work when merging changes. The compilation time measurements are done on a machine with Intel(R) Core(TM) i5-6500 CPU @ 3.20GHz (4 CPUs, 1 thread per core) and 8GB of memory.
The results for the version with no changes (3 times each, to see how random it can get):

time -p make -j 1 install
real 516.83
user 476.91
sys 39.67
real 518.56
user 477.19
sys 41.26
real 524.67
user 483.87
sys 40.47
time -p make -j 4 install
real 233.00
user 588.00
sys 51.27
real 178.04
user 569.57
sys 47.92
real 182.98
user 565.65
sys 46.99

[knopers8]

After the first round, which involved adding headers which before were used indirectly, deleting the unnecessary ones and using the correct inclusion type ("" or <>), these are the results:
time -p make -j 1 install
real 585.37
user 537.67
sys 46.30
time -p make -j 4 install
real 262.96
user 659.00
sys 58.25
This is clearly a downgrade, probably due to more diligent inclusions. Now, let's see how much we can improve by putting forward declarations where it is possible.
EDIT: These results are not reliable, the test setup was not stable enough, see below for test results on a server machine.

[knopers8]

After forward declarations and moving some function definitions to source files:
time -p make -j 1 install
real 578.93
user 525.55
sys 46.49
time -p make -j 4 install
real 221.72
user 600.49
sys 56.16
A bit better, but still worse than at the beginning. It would be good to confirm the results on a server machine, though as for now, it doesn't look promising.
EDIT: These results are not reliable, the test setup was not stable enough, see below for test results on a server machine.

[knopers8]

I tried it on aido2qc10 (Intel(R) Xeon(R) CPU E5-2640 v3 @ 2.60GHz, 32 CPUs, 128GB RAM):

master:
time -p make -j 1 install
real 530.61
user 489.13
sys 40.41
time -p make -j 4 install
real 163.05
user 515.02
sys 41.32
time -p make -j 32 install
real 68.43
user 792.75
sys 55.56

68d6d43 (deleting unnecessary headers and adding ones which were previously dangerously transient):

time -p make -j 1 install
real 516.40
user 475.64
sys 40.17
time -p make -j 4 install
real 157.60
user 504.50
sys 41.60
time -p make -j 32 install
real 65.02
user 772.59
sys 55.34

649d280 (forward declarations in the Framework, current progress):

time -p make -j 1 install
real 500.85
user 461.13
sys 38.93
time -p make -j 4 install
real 154.38
user 489.81
sys 39.15
time -p make -j 32 install
real 63.52
user 765.96
sys 53.57

Now the improvements are quite evident and coherent for different number of jobs specified, so I would blame the previous decreases on that test setup (a PC, IDE and browser open).

[knopers8]

7009614 (more forward declarations in Framework and header cleanup in tests):

time -p make -j 1 install
real 476.43
user 437.92
sys 37.96
time -p make -j 4 install
real 139.18
user 464.00
sys 39.18
time -p make -j 32 install
real 53.28
user 725.59
sys 51.64

[knopers8]

7198e59:

time -p make -j 1 install
real 474.55
user 435.70
sys 38.33
time -p make -j 4 install
real 139.30
user 462.48
sys 38.89
time -p make -j 32 install
real 53.65
user 719.47
sys 52.17

The influence of the changes in Modules was minimal.

[knopers8]

d9f2ef4:

time -p make -j 1 install
real 471.57
user 433.55
sys 37.33
time -p make -j 4 install
real 136.46
user 458.56
sys 39.35
time -p make -j 32 install
real 52.29
user 725.33
sys 51.87

A slight improvement by using = default ctors and dtors wherever possible.

[knopers8]

fef7fe1:

time -p make -j 1 install
real 468.86
user 438.69
sys 29.27
time -p make -j 4 install
real 131.24
user 450.72
sys 31.97
time -p make -j 32 install
real 50.97
user 707.96
sys 43.25

[knopers8]

Overall, we have 12% improvement for -j 1, 20% for -j 4 and 26% for -j 32 if we compare the "real" time.

Travis fails because of:

E: Package 'cmake' has no installation candidate
E: Package 'graphviz' has no installation candidate
apt-get.diagnostics
apt-get install failed
The command "sudo -E apt-get -yq --no-install-suggests --no-install-recommends $(travis_apt_get_options) install cmake doxygen doxygen-doc doxygen-latex doxygen-gui graphviz" failed and exited with 100 during .

I have applied clang-format to the PR, other builds are passing. Merging.