Remove generic AbstractArray fallbacks for position

[mtfishman]

Closes #16. This removes generic fallback definitions for the positions of eltype and ndims for AbstractArrays, which is safer since not all of them have eltype in the first position and ndims in the second position.

This was inspired by a bug I was hunting down while working on ITensor/QuantumOperatorDefinitions.jl#22, which was due to a BitArray using the generic position definition, which was incorrect because it doesn't have an eltype parameter since it is always Bool, and ndims is in the first position. In this PR, I added explicit correct definitions for BitArray.

@lkdvos @kmp5VT

I think this will be a good move. One issue this PR reminds me of is that right now default parameters are defined like:

default_type_parameters(::Type{<:Array}) = (Float64, 1)

i.e. they are defined in a list. This is ok, but I think it would be nicer if we could define them one at a time using the parameter names:

default_type_parameter(::Type{<:Array}, eltype) = Float64
default_type_parameter(::Type{<:Array}, ndims) = 1

I remember we tried that earlier on and it was tough to make type stable, but I think it would be good to revisit that (but not in this PR). (EDIT: Raised an issue here: #27.)

[codecov]

Codecov Report

Attention: Patch coverage is 75.00000% with 3 lines in your changes missing coverage. Please review.

Project coverage is 86.38%. Comparing base (61b969d) to head (6964e47).
Report is 5 commits behind head on main.

Files with missing lines	Patch %	Lines
src/base/abstractarray.jl	62.50%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #26      +/-   ##
==========================================
+ Coverage   84.86%   86.38%   +1.52%     
==========================================
  Files          10        9       -1     
  Lines         185      191       +6     
==========================================
+ Hits          157      165       +8     
+ Misses         28       26       -2     

Flag	Coverage Δ
docs	47.87% <58.33%> (+0.07%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lkdvos]

Definitely agree to get rid of that definition, as this clearly might silently give you wrong results.

Also agreed to having the option of specifying defaults one at a time, in a future PR.

[kmp5VT]

This makes sense to me, I had concerns with these definitions before. Thanks Matt!

[mtfishman]

@lkdvos any more comments?

[lkdvos]

Looks good to me!