fix: bring costs.txt inline with the actual protocol costs

[matklad]

forgot to do this in #4979

This doesn't affect any logic anywhere, just makes it easier to figure
out what our measured costs are.

[matklad]

CI failure seems unrelated:

[2021-10-26T12:12:03Z]
--
  | [2021-10-26T12:12:03Z] failures:
  | [2021-10-26T12:12:03Z]
  | [2021-10-26T12:12:03Z] ---- trie::split_state::tests::test_get_trie_items_for_part stdout ----
  | [2021-10-26T12:12:03Z] thread 'trie::split_state::tests::test_get_trie_items_for_part' panicked at 'part length 1 avg length 0', core/store/src/trie/split_state.rs:451:13
  | [2021-10-26T12:12:03Z] note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace
  | [2021-10-26T12:12:03Z]
  | [2021-10-26T12:12:03Z]
  | [2021-10-26T12:12:03Z] failures:
  | [2021-10-26T12:12:03Z]     trie::split_state::tests::test_get_trie_items_for_part

(cc @mzhangmzz ? split-state seems sharding related)

[stale]

This PR has been automatically marked as stale because it has not had recent activity in the 2 weeks.
It will be closed in 3 days if no further activity occurs.
Thank you for your contributions.

[mzhangmzz]

sorry I just saw this. @matklad that test was flaky but I fixed it. It should work if you merge with the latest master

[stale]

This PR has been automatically marked as stale because it has not had recent activity in the 2 weeks.
It will be closed in 3 days if no further activity occurs.
Thank you for your contributions.

[matklad]

Hopelessly obsolete, but let me tag @jakmeier here :)

The idea was that we need an easy reference for what the costs are today.