copy-tree does not copy float-vector

[knorth55]

I know copy-tree is for list, but copy-tree returns no error with float-vector and does not copy the vector.
It is better to warn copy-tree with float-vector.

1.irteusgl$ (setq a  (float-vector 1 2 3))
#f(1.0 2.0 3.0)
2.irteusgl$ (setq b (copy-tree a))
#f(1.0 2.0 3.0)
3.irteusgl$ (setf (elt b 0) 5)
5
4.irteusgl$ a
#f(5.0 2.0 3.0)
5.irteusgl$ b
#f(5.0 2.0 3.0)

[Affonso-Gui]

The implementation explicitly returns the non-cloned value when the argument is not a list, so this doesn't seem like some sort of upredicted side effect (i.e. designed for lists but that "accidentally" could handle other sequence types).
https://github.com/euslisp/EusLisp/blob/master/lisp/c/lists.c#L236

I am not sure if we should solve this by changing the implementation to also copy non-lists; add a warning; or just add more specific jmanual documentation. However, this is very similar to #361 , which promises destructive results but only delivers that on some cases.

[knorth55]

I think it is better to keep old behavior for keep backward compatibility, but we can warn the misusage or non-copied behavior.
so I open a PR #476

[Affonso-Gui]

(just for future reference I would like to add that in this case copy-seq does what you want)

irteusgl> (setq a (float-vector 1 2 3))
#f(1.0 2.0 3.0)
irteusgl> (setq b (copy-seq a))
#f(1.0 2.0 3.0)
irteusgl> (setf (elt b 0) 5)
5
irteusgl> a
#f(1.0 2.0 3.0)
irteusgl> b
#f(5.0 2.0 3.0)

irteusgl> (sys:address a)
94797007551080
irteusgl> (sys:address b)
94797007912976

EDIT: or copy-object if you need a deep copy

[knorth55]

copy-seq, which I use it for float-vector normally.
I misused copy-tree for float-vector, which returns non-cloned float-vector, and I spent several hours for debugging it.