Added init/deinit to BenchmarkResults and updated examples

[Bryysen]

This was done to circumvent #49, which is (most likely) caused by a bug in the compiler. Closes #49.

[Bryysen]

This was kind of a headache to get to the bottom of.. this solution didn't feel great but it's the one that requires the least amount of work for now. The alternative was of course to drop the buffered std-out writer all-together.

Calling getStdOut() inside a function seems to circumvent the issue, which is the reason for adding the BenchmarkResults.init function to do the initialization, instead of initializing its fields using default values.

[FObersteiner]

I guess this does it as a temporary work-around. In general, I think there's two things that should be on the agenda:

• clarify (by a use-case maybe) why there is a configurable writer instead of a hard-coded buffered writer to stdout
• revise benchmark setup and run methods. Thinking as a new user, I feel it can be confusing which part needs to initialized (why do I need to set up results and benchmark separately?) and which part needs to be de-initialized (why the results? why not the benchmark? why not the allocator?)

[hendriknielaender]

I guess this does it as a temporary work-around. In general, I think there's two things that should be on the agenda:

• clarify (by a use-case maybe) why there is a configurable writer instead of a hard-coded buffered writer to stdout
• revise benchmark setup and run methods. Thinking as a new user, I feel it can be confusing which part needs to initialized (why do I need to set up results and benchmark separately?) and which part needs to be de-initialized (why the results? why not the benchmark? why not the allocator?)

I also think the api could be leaner, something like this should do the job:

    var bench = Benchmark.init(std.heap.page_allocator, .{
        .iterations = 10,
    });
    defer bench.deinit();
    try bench.add("basic example", basicfn);
    try bench.add("sleep example", sleepfn);
    try bench.run();

But good points - i will open an issue for this to move the discussion. #52

Also from my side for now i think it does the job, feel free to merge it 👍