Consistent reads

[danielealbano]

This PR implements a mechanism to guarantee consistent reads adding support for read-only transactions and leveraging this new transaction type to synchronize the read and write operations.

A major advantage of this approach is that meanwhile before commands like MSET or MSETNX would have allowed a client to read a value of a sequence being set, breaking the ABI with the Redis interface (where MSET, MSETNX and all the other commands that operate on the same set of keys are atomic), now the new mechanism implemented directly into the transaction logic and leveraged internally by the hashtable, ensures that the reads will wait any write transaction in progress.

Meanwhile this is causing a performance penalty, the major advantage is that now cachegrand is much more ACID compliant (infact when used with the on-disk storage cachegrand is fully ACID compliant) providing transaction isolation.

Another major advantage is that now a lot of complexity implemented to allow reads with writes in progress can be dropped, which will lead to (a) an esamplification of the code and (b) some performance improvements.

These changes have room for improvement, the locks now might last for several seconds, depending on the type of operation being carried out, and there is no reason for a fiber to wait spinning, it makes much more sense to yield after some spinning to let other fibers to carry out their own operations.
Also, so far all the operation that required access to multiple keys were forced to use read-write transaction but for several it's not really required and they can switch to read-only transaction.

Because there is (plenty) of room for improvement, no benchmark is included to compare the performances.

[codecov]

Codecov Report

Attention: 7 lines in your changes are missing coverage. Please review.

Files	Coverage Δ
src/config.c	78.02% <ø> (ø)
src/data_structures/hashtable/mcmp/hashtable.c	84.62% <ø> (ø)
.../data_structures/hashtable/mcmp/hashtable_config.c	100.00% <ø> (ø)
...rc/data_structures/hashtable/mcmp/hashtable_data.c	81.48% <ø> (ø)
.../data_structures/hashtable/mcmp/hashtable_op_get.c	100.00% <100.00%> (ø)
...a_structures/hashtable/mcmp/hashtable_op_get_key.c	39.47% <ø> (ø)
...tures/hashtable/mcmp/hashtable_op_get_random_key.c	100.00% <ø> (ø)
...data_structures/hashtable/mcmp/hashtable_op_iter.c	93.33% <ø> (ø)
.../data_structures/hashtable/mcmp/hashtable_op_rmw.c	98.41% <ø> (ø)
.../data_structures/hashtable/mcmp/hashtable_op_set.c	100.00% <ø> (ø)
... and 90 more

... and 2 files with indirect coverage changes

📢 Thoughts on this report? Let us know!.