perf(shareReplay): use correct default paramter

[nename0]

Description:
shareReplay() uses a wrong default value for windowTime parameter since #3872. To disable the time function of the internal ReplaySubject the windowTime must be Number.POSITIVE_INFINITY and not Number.MAX_VALUE.
This makes shareReplay() faster because the ReplaySubject will not call scheduler.now(), when it is not needed.

This PR also adds a test, whichs checks if the ReplaySubject is in _infiniteTimeWindow mode, when no windowTime is given to shareReplay().

This will need merging with #4059 whatever comes first.

[coveralls]

Coverage remained the same at 96.798% when pulling 063a4a5 on nename0:fix-sharereplay into 6a475cc on ReactiveX:master.

[cartant]

LGTM.

sinon is used throughout the test suite. Could you use it to spy on now instead of the explicit function assignment?

const spy = sinon.spy(rxTestScheduler, 'now');
// ...
spy.restore();
expect(spy.called, 'ReplaySubject should not call ...').to.be.false;

[nename0]

I also added a test to ReplaySubject to get coverage back up.

[cartant]

Thanks for changing the test to use sinon. However, something appears to have happened to the alignment of the other marble diagrams in spec/operators/shareReplay-spec.ts. Do you have some sort of automated formatting enabled?

Regarding the coverage, it won't have been anything you did. The coverage seems to move around +/- 0.2% seemingly on a whim.

I think I'll have to look at the additional test when I'm less tired. I can't get my head around it, at the moment.

Thanks again for the PR.

[nename0]

Thanks for the fast reponse!

It's not automated just a habit of pressing the format document shorcut every once in a while, which I should not do in a foreign project.
About the coverage: This line only gets run when both bufferSize and windowTime are used. Previously there was no test with both paramters, so I added one.
But still a few more test should not hurt.
Also shareReplay-spec.ts doesn't test the windowTime param at all. This is still true in #4059