C++: Fix join-order in HttpStringLiteral charpred

[MathiasVP]

The char pred of HttpStringLiteral did great on almost every project on LGTM. It turns out the join-order wasn't the best possible one, however. This led to a single outlier on LGTM (specifically, this project: https://github.com/e-ago/bitcracker).

This PR fixes the join order.

On main (on https://github.com/e-ago/bitcracker):

Tuple counts for UseOfHttp::HttpStringLiteral#class#f#antijoin_rhs/2@d95908b5 after 45m50s:
  65980      ~0%     {3} r1 = JOIN quick_eval#f#shared WITH Expr::Expr::getValue_dispred#ff ON FIRST 2 OUTPUT Lhs.0 'arg0', Lhs.0 'arg0', Lhs.1 'arg1'
                      
  131968     ~1%     {3} r2 = JOIN quick_eval#f#shared#1 WITH #Expr::Expr::getParent_dispredPlus#bf ON FIRST 1 OUTPUT Rhs.1, Lhs.0 'arg0', Lhs.1 'arg1'
                      
  197948     ~0%     {3} r3 = r1 UNION r2
  4295033318 ~0%     {3} r4 = JOIN r3 WITH TaintTrackingUtil::localExprTaint#bf_10#join_rhs ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'arg0', Lhs.2 'arg1'
  4295033313 ~0%     {3} r5 = JOIN r4 WITH Literal::StringLiteral#f ON FIRST 1 OUTPUT Lhs.0, Lhs.1 'arg0', Lhs.2 'arg1'
  4295033313 ~0%     {4} r6 = JOIN r5 WITH Expr::Expr::getValue_dispred#ff ON FIRST 1 OUTPUT Lhs.1 'arg0', Lhs.2 'arg1', Rhs.1, "(?i)localhost(?:[:/?#].*)?|127\\.0\\.0\\.1(?:[:/?#].*)?|10(?:\\.[0-9]+){3}(?:[:/?#].*)?|172\\.16(?:\\.[0-9]+){2}(?:[:/?#].*)?|192.168(?:\\.[0-9]+){2}(?:[:/?#].*)?|\\[?0:0:0:0:0:0:0:1\\]?(?:[:/?#].*)?|\\[?::1\\]?(?:[:/?#].*)?"
  0          ~0%     {4} r7 = SELECT r6 ON In.2 matches_constant "(?i)localhost(?:[:/?#].*)?|127\.0\.0\.1(?:[:/?#].*)?|10(?:\.[0-9]+){3}(?:[:/?#].*)?|172\.16(?:\.[0-9]+){2}(?:[:/?#].*)?|192.168(?:\.[0-9]+){2}(?:[:/?#].*)?|\[?0:0:0:0:0:0:0:1\]?(?:[:/?#].*)?|\[?::1\]?(?:[:/?#].*)?"
  0          ~0%     {2} r8 = SCAN r7 OUTPUT In.0 'arg0', In.1 'arg1'
  return r8

On this PR:

Tuple counts for UseOfHttp::privateHostNameFlowsToExpr#f/1@8e3d0b6e after 122ms:
  65980 ~0%     {3} r1 = JOIN Literal::StringLiteral#f WITH Expr::Expr::getValue_dispred#ff ON FIRST 1 OUTPUT Lhs.0, Rhs.1, "(?i)localhost(?:[:/?#].*)?|127\\.0\\.0\\.1(?:[:/?#].*)?|10(?:\\.[0-9]+){3}(?:[:/?#].*)?|172\\.16(?:\\.[0-9]+){2}(?:[:/?#].*)?|192.168(?:\\.[0-9]+){2}(?:[:/?#].*)?|\\[?0:0:0:0:0:0:0:1\\]?(?:[:/?#].*)?|\\[?::1\\]?(?:[:/?#].*)?"
  0     ~0%     {3} r2 = SELECT r1 ON In.1 matches_constant "(?i)localhost(?:[:/?#].*)?|127\.0\.0\.1(?:[:/?#].*)?|10(?:\.[0-9]+){3}(?:[:/?#].*)?|172\.16(?:\.[0-9]+){2}(?:[:/?#].*)?|192.168(?:\.[0-9]+){2}(?:[:/?#].*)?|\[?0:0:0:0:0:0:0:1\]?(?:[:/?#].*)?|\[?::1\]?(?:[:/?#].*)?"
  0     ~0%     {1} r3 = SCAN r2 OUTPUT In.0
  0     ~0%     {1} r4 = JOIN r3 WITH DataFlowUtil::TExprNode#ff ON FIRST 1 OUTPUT Rhs.1
  0     ~0%     {1} r5 = JOIN r4 WITH TaintTrackingUtil::localTaint#bf ON FIRST 1 OUTPUT Rhs.1
  0     ~0%     {1} r6 = JOIN r5 WITH DataFlowUtil::TExprNode#ff_10#join_rhs ON FIRST 1 OUTPUT Rhs.1 'e'
  return r6
...
Inferred that UseOfHttp::HttpStringLiteral#class#f#antijoin_rhs/1@f2b613a7 is empty, due to UseOfHttp::privateHostNameFlowsToExpr#f/1@8e3d0b6e.

Well, that fixed it. But I'm not super happy with the validation of the performance yet 😂.

Here's another random project that shows that the join order is fine even when UseOfHttp::privateHostNameFlowsToExpr#f is non-empty (I picked php):

Tuple counts for UseOfHttp::privateHostNameFlowsToExpr#f/1@2b0c936l after 41ms:
  40478 ~0%     {3} r1 = JOIN Literal::StringLiteral#f WITH Expr::Expr::getValue_dispred#ff ON FIRST 1 OUTPUT Lhs.0, Rhs.1, "(?i)localhost(?:[:/?#].*)?|127\\.0\\.0\\.1(?:[:/?#].*)?|10(?:\\.[0-9]+){3}(?:[:/?#].*)?|172\\.16(?:\\.[0-9]+){2}(?:[:/?#].*)?|192.168(?:\\.[0-9]+){2}(?:[:/?#].*)?|\\[?0:0:0:0:0:0:0:1\\]?(?:[:/?#].*)?|\\[?::1\\]?(?:[:/?#].*)?"
  4     ~0%     {3} r2 = SELECT r1 ON In.1 matches_constant "(?i)localhost(?:[:/?#].*)?|127\.0\.0\.1(?:[:/?#].*)?|10(?:\.[0-9]+){3}(?:[:/?#].*)?|172\.16(?:\.[0-9]+){2}(?:[:/?#].*)?|192.168(?:\.[0-9]+){2}(?:[:/?#].*)?|\[?0:0:0:0:0:0:0:1\]?(?:[:/?#].*)?|\[?::1\]?(?:[:/?#].*)?"
  4     ~0%     {1} r3 = SCAN r2 OUTPUT In.0
  4     ~0%     {1} r4 = JOIN r3 WITH DataFlowUtil::TExprNode#ff ON FIRST 1 OUTPUT Rhs.1
  16    ~0%     {1} r5 = JOIN r4 WITH TaintTrackingUtil::localTaint#bf ON FIRST 1 OUTPUT Rhs.1
  16    ~6%     {1} r6 = JOIN r5 WITH DataFlowUtil::TExprNode#ff_10#join_rhs ON FIRST 1 OUTPUT Rhs.1 'e'
  return r6
...
Tuple counts for UseOfHttp::HttpStringLiteral#class#f#antijoin_rhs/1@237e621b after 11ms:
  40478  ~0%       {2} r1 = JOIN Literal::StringLiteral#f WITH Literal::StringLiteral#f ON FIRST 1 OUTPUT Rhs.0 'arg0', Lhs.0 'arg0'
  40478  ~0%       {2} r2 = JOIN r1 WITH Literal::StringLiteral#f ON FIRST 1 OUTPUT Lhs.0, Lhs.1 'arg0'
                      
  4      ~0%       {1} r3 = JOIN r2 WITH UseOfHttp::privateHostNameFlowsToExpr#f ON FIRST 1 OUTPUT Lhs.1 'arg0'
                      
  127947 ~2%       {2} r4 = JOIN r2 WITH #Expr::Expr::getParent_dispredPlus#bf ON FIRST 1 OUTPUT Rhs.1, Lhs.1 'arg0'
  4      ~100%     {1} r5 = JOIN r4 WITH UseOfHttp::privateHostNameFlowsToExpr#f ON FIRST 1 OUTPUT Lhs.1 'arg0'
                      
  8      ~100%     {1} r6 = r3 UNION r5
  return r6

[geoffw0]

Thanks for fixing this while I was away!