add trackers for named reward hacks

[TensorTemplar]

No description provided.

[github-actions]

📊 Slopometry QPE Report

QPE Score: 0.2196

Metric	Value	Description
MI (normalized)	0.262	Maintainability Index / 100
Smell Penalty	0.239	Weighted code smell deduction
Adjusted Quality	0.220	MI × (1 - smell_penalty) + bonuses

Code Smell Breakdown

Smell	Count
orphan comment	239
inline import	220
dict get with default	73
swallowed exception	30
single method class	25
passthrough wrapper	15
hasattr getattr	14
untracked todo	7
test skip	5
type ignore	2

Higher QPE = better quality

[github-actions]

📈 Slopometry Impact Report

Impact: NEUTRAL (score: -0.135)

Metric	Delta	Description
QPE	-0.0082	Quality-Per-Effort change
MI	-38.611	Maintainability Index change
CC	+282.000	Cyclomatic Complexity change
Effort	+2971377.3	Halstead Effort change

	Count
Changed files	19
Blind spots	4

Smell Advantage Breakdown

Smell	Baseline	Current	Weighted Delta
swallowed exception	27	30	+0.4500
test skip	3	5	+0.2000
dict get with default	71	73	+0.1000
inline import	211	220	+0.0900
orphan comment	233	239	+0.0600
single method class	24	25	+0.0500

Source: Previous commit (dd8f045 vs 7fc1e63)