[Issue-126] fix several issues with animate - #131

## Summary

`tranimate(T, dim=1.5)` — using the scalar-shorthand `dim` convention
established by `plotvol2`/`plotvol3` elsewhere in this codebase —
currently
crashes with:

```
TypeError: spatialmath.base.animate.Animate() got multiple values for keyword argument 'dim'
```

Two compounding bugs, both fixed here:

1. `tranimate()` popped `"dims"` (plural) out of `kwargs` before
forwarding
   to `Animate(dim=..., **kwargs)`, but every other function in this
codebase (`plotvol2`, `plotvol3`, `Animate.__init__` itself) names the
   parameter `dim` (singular). A caller correctly using `dim=` per that
   convention never got popped, so it stayed in `**kwargs` and collided
   with the explicit `dim=dim` already being passed — hence "multiple
   values for keyword argument 'dim'".
2. Even with the right keyword, `Animate.__init__` had its own
hand-rolled
   length check that only accepted a 2- or 6-element `dim`, silently
   dropping the scalar-shorthand convention (`A` → `[-A, A]` on every
   axis) that `plotvol3` already supports via the shared `expand_dims()`
   helper — `Animate` just never called it.

## Fix

- `tranimate()` now accepts `dim`, with `dims` kept as a back-compat
alias
  for existing callers following the old docstring examples.
- `Animate.__init__` now calls `expand_dims(dim, nd=3)` (the same helper
`plotvol3` uses) instead of its own incomplete validation, so scalar and
  2-element shorthand work consistently everywhere.

## Regression context

This isn't a long-standing issue — it's a regression from #131 (`Jien
Cao`,
2024-07-29, fixing #126), which *introduced* both the `dims` rename and
the
new strict length check while trying to "expose `ax` and `dims`
arguments
to animate(...), to be consistent with other APIs." Before that commit,
`tranimate` passed `**kwargs` straight through to `Animate`, and scalar
`dim` worked correctly via `Animate`'s existing `plotvol3` delegation.

## Test plan

- [x] Verified directly: `Animate(dim=1.5)`, `Animate(dim=[0,5])`,
      `tranimate(R, dim=1.5, movie=True)`, and `tranimate(R, dims=[0,5],
      movie=True)` (back-compat alias) all construct/run correctly
- [x] Full existing test suite: 342 passed, no regressions