CSHARP-979 Improve data reading performance by inlining hot methods in RecyclableMemoryStream

[marcin-krystianc]

As explained in the Jira ticket, these methods are executed a lot of times. After inlining them, I can see improvements of 3-5% in reading data. I know that it is a micro-optimisation that is rarely worth doing. However, in this case, it gives non-negligible performance improvements.