fix: Validate and normalize inet test values (#2205)

#### Summary

The inet type, uses the `net.ParseCIDR` function which when returning the network, actually drop some of the details of the CIDR string that is passed in...

```
It returns the IP address and the network implied by the IP and prefix length. For example, ParseCIDR("192.0.2.1/24") returns the IP address 192.0.2.1 and the network 192.0.2.0/24.
```

This fix ensures that all inet values will be unchanged if they go through a read --> write --> read (like we do in the filetypes tests)


This issue was discovered in the filetypes tests where a value was written to a parquet file then read back into an arrow record where the inet builder normalized the data and when compared to the original record a difference was detected

Author: bbernays

schema/testdata.go

@@ -5,6 +5,7 @@ import (
 	"encoding/base64"
 	"fmt"
 	"math/rand"
+	"net"
 	"strconv"
 	"strings"
 	"time"
@@ -316,9 +317,11 @@ func (tg TestDataGenerator) getExampleJSON(colName string, dataType arrow.DataTy
 
 		// Generate a CIDR prefix length between 8 and 30
 		cidr := 8 + rnd.Intn(23)
+		input := fmt.Sprintf(`%d.%d.%d.%d/%d`, ip[0], ip[1], ip[2], ip[3], cidr)
+		_, data, _ := net.ParseCIDR(input)
 
 		// Format as an IP address with CIDR notation
-		return fmt.Sprintf(`"%d.%d.%d.%d/%d"`, ip[0], ip[1], ip[2], ip[3], cidr)
+		return fmt.Sprintf(`"%s"`, data.String())
 	}
 	if arrow.TypeEqual(dataType, types.ExtensionTypes.MAC) {
 		mac := make([]byte, 6)