High CPU from append: frequent runtime.growslice due to missing slice preallocation

## Describe the bug
- The `Append` built-in uses excessive CPU due to frequent runtime.growslice calls.

## Steps to reproduce
1. Run an INSERT that writes a large number of rows.
2. This triggers multiple append operations that allocate new underlying arrays because the current slice has no predefined capacity.
3. Capture a Go CPU profile and inspect the hot spots for append and runtime.growslice.

## Expected behavior
- Provide a way to preallocate slice capacity (or otherwise reduce allocations) so that calls to runtime.growslice are minimized or eliminated.

## Code example
- The current column type does not allow setting the initial capacity of the slice:
```go
type ColDateTime64 struct {
Data []DateTime64
Location *time.Location
Precision Precision
PrecisionSet bool
}
```

Start with the ColDateTime64 type shown in the issue and trace how its Data slice is populated during large INSERT operations. Use a Go CPU profile to confirm runtime.growslice activity, then define and implement a capacity-preallocation approach that reduces those allocations and verify the profiling result.

Written by the indexing model from the issue text.