RFR: 8253286: Use expand_exact() instead of expand_at() for fixed requests in G1

[Albert Mingkun Yang]

On Fri, 18 Sep 2020 13:40:42 GMT, Stefan Johansson <sjohanss at openjdk.org> wrote:

> we should never fail this expand

I think I am missing some background knowledge. I assume `expand_exact()`, in generally, could fail, but for args,
`curr, 1, NULL` here, it will not fail, because ... The "because" part should be in the comments right above the call.

As for the post-assert, `_regions.get_by_index(curr) != NULL && is_available(curr)`, I think they are already covered
in `expand_exact`.

I like `at(curr)->is_free()` better, since the current function, `find_highest_free`, should return a region (by index)
that's free.

-------------

PR: https://git.openjdk.java.net/jdk/pull/235