List reference tables readable by the authenticated organization.

Returns your organization's tables plus published Nebulock-curated ones. Retired
(soft-deleted) tables are never returned.

items is omitted by default -- loading every row of every table to render a
page is the cost this shape avoids. Two counts describe the size without them:
entry_count is how many (field_name, value) entries the table carries, and
item_count is how many rows hold them. They differ whenever a row covers more than
one field, which is every multi-column CSV upload.

include_items=true loads them anyway, in one query for the whole page. Ask for it
knowingly: a table holds up to 2,500 entries, so a full page of large tables is a very
large response. Pair it with a smaller limit, or fetch the one table you need.

slug matches exactly, which search does not: a search for browsers also
returns a table that merely mentions it in a description, so it cannot answer "which
table does ref_table('browsers') mean". Case-sensitive, and stored slugs are always
lowercase, so a capitalised value matches nothing rather than guessing. Both filters
may be combined.

A slug lookup can return two tables: your own and a published Nebulock-curated one
may share an unprefixed name, and you can read both. Your own sorts first, so
data[0] is the table ref_table('<slug>') would expand to -- the same precedence
expansion applies. That ordering outranks order_by for a slug lookup, and only for
one; add &type=customer or &type=nebulock to see just one side.