fix: include `PartitionHashId` in size estimations (#8153)

As for the other types: size estimations are conservative, so we assume
the value behind the `Arc` is owned by the estimating party.

data_types/src/lib.rs

@@ -614,7 +614,13 @@ impl ParquetFile {
 
     /// Estimate the memory consumption of this object and its contents
     pub fn size(&self) -> usize {
-        std::mem::size_of_val(self) + self.column_set.size()
+        std::mem::size_of_val(self)
+            + self
+                .partition_hash_id
+                .as_ref()
+                .map(|id| id.size() - std::mem::size_of_val(id))
+                .unwrap_or_default()
+            + self.column_set.size()
             - std::mem::size_of_val(&self.column_set)
     }
 

data_types/src/partition.rs

@@ -19,6 +19,18 @@ pub enum TransitionPartitionId {
     Deterministic(PartitionHashId),
 }
 
+impl TransitionPartitionId {
+    /// Size in bytes including `self`.
+    pub fn size(&self) -> usize {
+        match self {
+            Self::Deprecated(_) => std::mem::size_of::<Self>(),
+            Self::Deterministic(id) => {
+                std::mem::size_of::<Self>() + id.size() - std::mem::size_of_val(id)
+            }
+        }
+    }
+}
+
 impl std::fmt::Display for TransitionPartitionId {
     fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
         match self {
@@ -216,6 +228,11 @@ impl PartitionHashId {
     pub fn as_bytes(&self) -> &[u8] {
         self.0.as_ref()
     }
+
+    /// Size in bytes including `Self`.
+    pub fn size(&self) -> usize {
+        std::mem::size_of::<Self>() + self.0.len()
+    }
 }
 
 impl<'q> sqlx::encode::Encode<'q, sqlx::Postgres> for &'q PartitionHashId {

querier/src/cache/parquet_file.rs

@@ -361,8 +361,8 @@ mod tests {
         partition.create_parquet_file(builder).await;
         let table_id = table.table.id;
 
-        let single_file_size = 208;
-        let two_file_size = 384;
+        let single_file_size = 240;
+        let two_file_size = 448;
         assert!(single_file_size < two_file_size);
 
         let cache = make_cache(&catalog);