If the number of sync requests is big enough, the palloc() call in
AbsorbSyncRequests() will attempt to allocate more than 1 GB of memory,
resulting in failure.  This can lead to an infinite loop in the checkpointer
process, as it repeatedly fails to absorb the pending requests.

This commit introduces the following changes to cope with this problem:
 1. Turn pending checkpointer requests array in shared memory into a bounded
    ring buffer.
 2. Limit maximum ring buffer size to 10M items.
 3. Make AbsorbSyncRequests() process requests incrementally in 10K batches.

Even #2 makes the whole queue size fit the maximum palloc() size of 1 GB.
of continuous lock holding.

This commit is for master only.  Simpler fix, which just limits a request
queue size to 10M, will be backpatched.

Reported-by: Ekaterina Sokolova <e.sokolova@postgrespro.ru>
Discussion: https://postgr.es/m/db4534f83a22a29ab5ee2566ad86ca92%40postgrespro.ru
Author: Maxim Orlov <orlovmg@gmail.com>
Co-authored-by:  Xuneng Zhou <xunengzhou@gmail.com>
Reviewed-by: Andres Freund <andres@anarazel.de>
Reviewed-by: Heikki Linnakangas <hlinnaka@iki.fi>
Reviewed-by: Alexander Korotkov <aekorotkov@gmail.com>

There've been a few complaints that it can be overly difficult to figure
out why the planner picked a Memoize plan.  To help address that, here we
adjust the EXPLAIN output to display the following additional details:

1) The estimated number of cache entries that can be stored at once
2) The estimated number of unique lookup keys that we expect to see
3) The number of lookups we expect
4) The estimated hit ratio

Technically postgres#4 can be calculated using #1, #2 and postgres#3, but it's not a
particularly obvious calculation, so we opt to display it explicitly.
The original patch by Lukas Fittl only displayed the hit ratio, but
there was a fear that might lead to more questions about how that was
calculated.  The idea with displaying all 4 is to be transparent which
may allow queries to be tuned more easily.  For example, if #2 isn't
correct then maybe extended statistics or a manual n_distinct estimate can
be used to help fix poor plan choices.

Author: Ilia Evdokimov <ilya.evdokimov@tantorlabs.com>
Author: Lukas Fittl <lukas@fittl.com>
Reviewed-by: David Rowley <dgrowleyml@gmail.com>
Reviewed-by: Andrei Lepikhov <lepihov@gmail.com>
Reviewed-by: Robert Haas <robertmhaas@gmail.com>
Discussion: https://postgr.es/m/CAP53Pky29GWAVVk3oBgKBDqhND0BRBN6yTPeguV_qSivFL5N_g%40mail.gmail.com

truncate_useless_pathkeys() seems to have neglected to account for
PathKeys that might be useful for WindowClause evaluation.  Modify it so
that it properly accounts for that.

Making this work required adjusting two things:

1. Change from checking query_pathkeys to check sort_pathkeys instead.
2. Add explicit check for window_pathkeys

For #1, query_pathkeys gets set in standard_qp_callback() according to the
sort order requirements for the first operation to be applied after the
join planner is finished, so this changes depending on which upper
planner operations a particular query needs.  If the query has window
functions and no GROUP BY, then query_pathkeys gets set to
window_pathkeys.  Before this change, this meant PathKeys useful for the
ORDER BY were not accounted for in queries with window functions.

Because of #1, #2 is now required so that we explicitly check to ensure
we don't truncate away PathKeys useful for window functions.

Author: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/CAApHDvrj3HTKmXoLMbUjTO=_MNMxM=cnuCSyBKidAVibmYPnrg@mail.gmail.com