SERVER-101933's investigation highlighted some current limitations with internal write concerns in sharded clusters.

We have a bunch of internal write concern timeouts that are used to override user-provided wtimeout for many operations. In particular, those 3 are widely used and often override user provided write concerns for DDL/administrative commands:

    static constexpr Seconds kWriteConcernTimeoutMigration{30};
    static constexpr Seconds kWriteConcernTimeoutSharding{60};
    static constexpr Seconds kWriteConcernTimeoutUserCommand{60};

In order to properly tackle SERVER-101933, refactor the usages and centralize/deduplicate write concern declarations based on those wtimeout.