bug: do_delete_sanity_check panic due to inconsistent delete

[yezizp2012]

Describe the bug

   0: rust_begin_unwind
             at /rustc/bdb07a8ec8e77aa10fb84fae1d4ff71c21180bb4/library/std/src/panicking.rs:575:5
   1: core::panicking::panic_fmt
             at /rustc/bdb07a8ec8e77aa10fb84fae1d4ff71c21180bb4/library/core/src/panicking.rs:64:14
   2: risingwave_stream::common::table::state_table::StateTable<S>::do_delete_sanity_check::{{closure}}
             at ./src/stream/src/common/table/state_table.rs:807:13
   3: risingwave_stream::common::table::state_table::StateTable<S>::batch_write_rows::{{closure}}
             at ./src/stream/src/common/table/state_table.rs:714:70
   4: risingwave_stream::common::table::state_table::StateTable<S>::commit::{{closure}}
             at ./src/stream/src/common/table/state_table.rs:641:57
   5: risingwave_stream::executor::hash_agg::HashAggExecutor<K,S>::flush_data::{{closure}}::{{closure}}::{{closure}}
             at ./src/stream/src/executor/hash_agg.rs:492:42

To Reproduce

1. start a 3cn cluster with recovery enabled
2. create nexmark_q5 and nexmark_q7
3. after running for a while, kill one of CNs, then two other nodes panicked.

Expected behavior

No response

Additional context

No response

[yezizp2012]

Looks like it will be fixed in #7082 .