Conditions that are outside of PK are marked as `unknown` in `KeyCondition`,
so it's safe to allow them, as long as they are always combined by `AND`.

Existing code that looks up marks that match the query has a pathological
case, when most of the part does in fact match the query.

The code works by recursively splitting a part into ranges and then discarding
the ranges that definitely do not match the query, based on primary key.

The problem is that it requires visiting every mark that matches the query,
making the complexity of this sort of look up O(n).

For queries that match exact range on the primary key, we can find
both left and right parts of the range with O(log 2) complexity.

This change implements exactly that.

To engage this optimization, the query must:

* Have a prefix list of the primary key.
* Have only range or single set element constraints for columns.
* Have only AND as a boolean operator.

Consider a table with `(service, timestamp)` as the primary key.

The following conditions will be optimized:

* `service = 'foo'`
* `service = 'foo' and timestamp >= now() - 3600`
* `service in ('foo')`
* `service in ('foo') and timestamp >= now() - 3600 and timestamp <= now`

The following will fall back to previous lookup algorithm:

* `timestamp >= now() - 3600`
* `service in ('foo', 'bar') and timestamp >= now() - 3600`
* `service = 'foo'`

Note that the optimization won't engage when PK has a range expression
followed by a point expression, since in that case the range is not continuous.

Trace query logging provides the following messages types of messages,
each representing a different kind of PK usage for a part:

```
Used optimized inclusion search over index for part 20200711_5710108_5710108_0 with 9 steps
Used generic exclusion search over index for part 20200711_5710118_5710228_5 with 1495 steps
Not using index on part 20200710_5710473_5710473_0
```

Number of steps translates to computational complexity.

Here's a comparison for before and after for a query over 24h of data:

```
Read 4562944 rows, 148.05 MiB in 45.19249672 sec.,   100966 rows/sec.,   3.28 MiB/sec.
Read 4183040 rows, 135.78 MiB in 0.196279627 sec., 21311636 rows/sec., 691.75 MiB/sec.
```

This is especially useful for queries that read data in order
and terminate early to return "last X things" matching a query.

See #11564 for more thoughts on this.

Avoid exception when negative or floating point constant is used in WHERE condition for indexed tables #11905

* Integrated CachingAllocator into MarkCache

* fixed build errors

* reset func hotfix

* upd: Fixing build

* updated submodules links

* fix 2

* updating grabber allocator proto

* updating lost work

* updating CMake to use concepts

* some other changes to get it building (integration into MarkCache)

* further integration into caches

* updated Async metrics, fixed some build errors

* and some other errors revealing

* added perfect forwarding to some functions

* fix: forward template

* fix: constexpr modifier

* fix: FakePODAllocator missing member func

* updated PODArray constructor taking alloc params

* fix: PODArray overload with n restored

* fix: FakePODAlloc duplicating alloc() func

* added constexpr variable for alloc_tag_t

* split cache values by allocators, provided updates

* fix: memcpy

* fix: constexpr modifier

* fix: noexcept modifier

* fix: alloc_tag_t for PODArray constructor

* fix: PODArray copy ctor with different alloc

* fix: resize() signature

* updating to lastest working master

* syncing with 273267

* first draft version

* fix: update Searcher to case-insensitive

* added ILIKE test

* fixed style errors, updated test, split like and ilike,  added notILike

* replaced inconsistent comments

* fixed show tables ilike

* updated missing test cases

* regenerated ya.make

* Update 01355_ilike.sql
Co-authored-by: Nmyrrc <me-clickhouse@myrrec.space>
Co-authored-by: Nalexey-milovidov <milovidov@yandex-team.ru>

Possible approach for fixing #10574

The problem is that prepared sets are built correctly, it is a hash map of key -> set
where key is a hash of AST and list of data types (when we a list of
tuples of literals).

However, when the key is built from the index to try and find if there
exists a prepared set that would match it looks for data types of the
primary key (see how data_types is populated) because the primary key
has only one field (v in my example) it can not find the prepared set.

The patch looks for any prepared indexes where data types match for the
subset of fields found in primary key, we are not interested in other
fields anyway for the purpose of primary key pruning.

* More LOGICAL_ERROR
* Proper cloning of some Asts
* Field::safeGet for user-supplied values

* Move some code outside dbms/src folder
* Fix paths