[Algorithm] Fix inverted trailer check condition in ForwardParser

[niteshg97]

Problem

ForwardParser::parse() in Algorithm/include/Algorithm/Parser.h contains
an inverted condition for trailer handling:

// Before (buggy)
if (tailOffset > 0) {        // tailOffset>0 means a TrailerType IS defined
    entry.trailer = nullptr;  // silently nulls it out — callback never called
} else {                      // else runs when there is NO trailer (void)
    ...CheckTrailer(entry, checkTrailer)...
}

tailOffset = typesize<TrailerType>::size, so it is > 0 precisely when
a real TrailerType is specified. The condition is completely inverted:

• When a TrailerType exists (tailOffset > 0), entry.trailer is set
  to nullptr and the user-supplied checkTrailer callback is never invoked.
• When TrailerType is void (tailOffset == 0), the dead else-branch runs
  a no-op CheckTrailer specialisation that always returns true.

Consequence: frames with corrupt or invalid trailers are silently accepted
without any validation. No crash occurs, making this bug invisible without
dedicated tests.

Why existing tests did not catch this

test_forwardparser_header_and_trailer provides a checkTrailer lambda but
never asserts it was actually called — it only checks payload bytes. The tests
passed even though the callback was completely bypassed.

Fix

Invert the condition (tailOffset > 0 → tailOffset == 0):

// After (fixed)
if (tailOffset == 0) {
    // no trailer type defined, nothing to extract
    entry.trailer = nullptr;
} else {
    auto trailerStart = buffer + position + frameSize - tailOffset;
    entry.trailer = reinterpret_cast<const TrailerType*>(trailerStart);
    if (!CheckTrailer(entry, checkTrailer)) {
        break;
    }
}

Tests

• Updated test_forwardparser_header_and_trailer:

  • Counts checkTrailer invocations and asserts it is called once per frame.
  • Asserts frames[i].trailer != nullptr and correct flags fields are extracted.

• Added test_forwardparser_trailer_check_rejection:

  • Provides a checkTrailer that rejects the second frame.
  • Asserts result == -1 (format error) and that insert is never called.

All existing tests (void-trailer, no-frames, format-error, reverse-parser)
are unaffected — they use TrailerType = void (tailOffset == 0) and are
handled by the unchanged if-branch.

[sawenzel]

Thanks for the patch, the fix is valid.

However, before merging we checked how ForwardParser is actually used, and it turns out the entire Algorithm/Parser.h parser utility (ForwardParser/ReverseParser), along with the related TableView.h, O2FormatParser.h and PageParser.h, has no callers anywhere in AliceO2 or O2Physics — the only references are the module's own unit tests. So this is a latent bug in code that is never instantiated.

Rather than fix a code path nobody exercises, I see an opportunity for cleanup and to remove the unused subsystem instead (see #15558). I therefore prefer to close this PR. Thanks again for spotting the logic error — the observation was valid, the underlying code just no longer has a place in the tree.