[ntuple] Add GetView<void> with type name string or std::type_info argument

[enirolf]

Enables passing a type name string to GetView<void> to signal that the pointer has a different underlying type than the on-disk field's type. The field used by RNTupleView will be constructed with this type, in turn benefitting from RNTuples built-in type- and bounds-checking.

To illustrate further, assuming myInt is a std::uint64_t on-disk:

std::int32_t myInt;
void* myIntPtr = &myInt;
auto myIntView = reader->GetView<void>("myInt", myIntPtr);
myIntView(0);
doSomething(myInt);

GetView<void> will currently always reconstruct the field used for loading the values from the on-disk descriptor. However, in the example above this can lead to undefined behavior when the actual values on disk don't fit in std::int32_t. The change proposed in this PR enables passing the type name string to GetView<void>, which is then used for the construction of the field instead, leveraging RNTuple's internal type- and bounds-checking:

std::int32_t myInt;
void* myIntPtr = &myInt;
auto myIntView = reader->GetView<void>("myInt", myIntPtr, "std::int32_t");
myIntView(0);
doSomething(myInt);

This feature was requested by and discussed with ATLAS.

[silverweed]

LGTM, just one question that probably doesn't belong to this PR

[github-actions]

Test Results

    17 files      17 suites   4d 22h 16m 20s ⏱️
 2 688 tests  2 688 ✅ 0 💤 0 ❌
45 074 runs  45 074 ✅ 0 💤 0 ❌

Results for commit 1d32f22.

♻️ This comment has been updated with latest results.

[hahnjo]

LG, thanks for the clear implementation!

[jblomer]

Many thanks! LGTM!

[vepadulano]

Thank you! But see comment before merging

[hahnjo]

As discussed on Friday, I removed the shared_ptr overload and un-templated GetView instead of the static_assert that T = void

[vepadulano]

Changed as discussed, LGTM