Client reset with recovery fixes

[ironage]

In stress testing the recovery code paths for https://jira.mongodb.org/browse/HELP-48043 I found a few errors.

1. We should not resurrect objects that have been deleted by the server when copying local lists with links in them. This could lead to some weird/unexpected state after a recovery, but I wasn't able to create any bad changesets this way.
   Related to Re-create all objects before updating their properties during client_reset::transfer_group() #6291

2. I also trace the problem reported in the forums by a user based on the exception message "A fatal error occured during client reset: 'Requested index 2 calling get() on set 'anotherschema.arrayfield' when max is 1' (2)." This turned out to be a problem with the LnkSet implementation of sort/distinct. The effect of returning duplicate indices for the client reset code is that the logic keeping track of indexes there breaks. I think the client reset code was the only place calling this (in core at least) so I don't think bug had too wide an impact.

☑️ ToDos

• 📝 Changelog update
• 🚦 Tests (or not relevant)
• C-API, if public C++ API changed.

[coveralls-official]

Pull Request Test Coverage Report for Build james.stone_426

• 303 of 317 (95.58%) changed or added relevant lines in 8 files are covered.
• 57 unchanged lines in 7 files lost coverage.
• Overall coverage increased (+0.006%) to 91.677%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
src/realm/object_converter.cpp	51	53	96.23%
test/object-store/collection_fixtures.hpp	12	15	80.0%
test/object-store/sync/client_reset.cpp	199	208	95.67%

Files with Coverage Reduction	New Missed Lines	%
test/test_table.cpp	1	99.64%
src/realm/sync/noinst/server/server_history.cpp	2	67.9%
src/realm/sync/network/network.cpp	3	89.65%
test/util/compare_groups.cpp	3	54.95%
src/realm/bplustree.cpp	7	75.72%
src/realm/sync/noinst/client_impl_base.cpp	12	85.0%
test/fuzz_group.cpp	29	50.0%

Totals
Change from base Build 1857:	0.006%
Covered Lines:	231579
Relevant Lines:	252603

---

💛 - Coveralls

[danieltabacaru]

👍

[kraenhansen]

@ironage this line in the changelog is missing a "since" entry:

Calling sort() or distinct() on a LnkSet that had unresolved links in it would produce duplicate indices.

Should I assume this was also since v11.16.0?

[ironage]

@kraenhansen good catch, that fix is relevant since the introduction of the Set datatype in v10.2.0