Feature: monitor cache-related perf events for L1D & LLC accesses/misses

[dssgabriel]

This PR adds support for monitoring L1D and LLC cache accesses & misses using Linux perf events.

Changelog:

• added cache events to the PerfCountSet and Measure data types;
• added cache events to the list of monitored hardware counters in the PerformanceCounters ctor:
  • needed to use the private monitor method because cache events require special configuration and would fail to compile due to -Wconversion and -Werror being enabled.
• added cache events to the benchmark's output;
• added cache events to the Mustache-like templates;
• changed miss% column to bra miss% for the sake of explicitness.

Sample output of this new feature:

| relative |                s/op |                op/s |    err% |          ins/op |          cyc/op |    IPC |         bra/op |  bra miss% |     L1D$ ref/op | L1D$ miss% |      LL$ ref/op |  LL$ miss% |     total | serial
|---------:|--------------------:|--------------------:|--------:|----------------:|----------------:|-------:|---------------:|-----------:|----------------:|-----------:|----------------:|-----------:|----------:|:-------
|   100.0% |                0.24 |                4.12 |    0.3% |  439,623,472.00 |  585,519,984.00 |  0.751 |  17,564,921.00 |       1.5% |  268,438,282.00 |     50.24% |  119,018,261.00 |      0.02% |      2.69 | `GEMM w/ standard layout`
|   169.8% |                0.14 |                7.00 |    0.3% |2,086,674,645.00 |  344,725,668.00 |  6.053 |  67,896,469.00 |       0.4% |  269,224,102.00 |     67.12% |    3,846,202.00 |      0.06% |      1.57 | `GEMM w/ tiled layout`

Rendered as:

relative	s/op	op/s	err%	ins/op	cyc/op	IPC	bra/op	bra miss%	L1D$ ref/op	L1D$ miss%	LL$ ref/op	LL$ miss%	total	serial
100.0%	0.24	4.12	0.3%	439,623,472.00	585,519,984.00	0.751	17,564,921.00	1.5%	268,438,282.00	50.24%	119,018,261.00	0.02%	2.69	GEMM w/ standard layout
169.8%	0.14	7.00	0.3%	2,086,674,645.00	344,725,668.00	6.053	67,896,469.00	0.4%	269,224,102.00	67.12%	3,846,202.00	0.06%	1.57	GEMM w/ tiled layout

[dssgabriel]

I'm not sure which test is failing and why it does.
I may have missed something for handling Mustache-like templates, do you have any pointers that I can look into @martinus? Thanks in advance!