Range for loop for statements

[Alvov1]

Closes #181

Adds a nested RowIterator class to Statement that enables range-based for loops over query results:

SQLite::Statement query(db, "SELECT id, name FROM test");
for (SQLite::Statement& row : query)
{
    std::cout << row.getColumn(0).getInt() << "\n";
}

Notes:

• begin() calls reset() automatically, so re-iterating the same Statement always starts from the beginning
• Bindings set before the loop are preserved across reset()
• Empty result sets are handled correctly (loop body never executes)
• RowIterator satisfies std::input_iterator_tag with full iterator traits

[SRombauts]

Thanks, all three earlier points are addressed. Two small follow-ups below, both non-blocking.

[SRombauts]

The implementation is fine, but the doc overstates it. It compares the Statement*, not the row, so two live iterators over the same statement always compare equal regardless of position. That is correct for a single-pass stream iterator (it is how std::istream_iterator works, and your @warning already rules out two live iterators). Only the wording needs to match:

/// Return true when both iterators refer to the same statement, or are both the end sentinel.

[SRombauts]

Worth flagging: the copy this returns shares the one live cursor, so *it++ gives the row after the increment, not before. This is inherent, not a bug to chase down. operator* hands back the single Statement&, Statement is move-only (Statement.h:79), and sqlite overwrites the row data on the next step, so the returned copy cannot hold the old row. std::istream_iterator gets away with *it++ because it caches its value by copy; here the value is a live cursor, so we cannot.

Caching the row would mean copying every column into the iterator, which is not worth it. The honest signal for a single-pass iterator is to return void from post-increment:

void RowIterator::operator++(int) { ++*this; }

That still satisfies std::input_iterator (weakly_incrementable only requires i++ to be a valid expression, with no constraint on its return type), and it turns auto old = *it++; into a compile error instead of a silent wrong-row read. If you would rather keep the conventional copy-returning form, add a @warning saying the iterator is single-pass and *it++ reflects the advanced row.

[Alvov1]

You're correct, a good catch
Returning a void, and producing a compile error is much safer than making a warning that could be missed

[SRombauts]

Hey @Alvov1, thanks a lot for your contribution!
I was going to approve and merge it as-is, but I discovered the subtleties of the operation++() return value (see review comment above).
It's not a big deal, but I'd like to know what you think of it: I would just return void if it was me, but you may have other ideas on how you would use it yourself
Cheers!
Sébastien

[SRombauts]

One more thought, and only if you are up for it: this feature is genuinely nice, and right now the only place it shows up is the tests. Would you be open to adding a small bit of documentation so people actually discover it? The natural spot is the query loop snippet in README.md (around line 341), where the classic while (query.executeStep()) example lives; showing the range-based for equivalent right next to it would do the job. A short variant in examples/example1/main.cpp would also work and has the bonus of being compiled by CI.

No worries if you would rather I add it after merging, just let me know which you prefer.

[Alvov1]

Hi! I'll make it in a couple of days. Didn't forget